There are 4 tests which can't currently be moved:
 - DownloadManagerTest explicitly uses the main looper which isn't easily
   supported because the test runs on this thread.
 - ContentDataSourceTest uses an AssetFileDescriptor which wraps a
   ParcelFileDescriptor. It seems Robolectric doesn't correctly forward the
   inner wrapped file descriptor leading to NPE.
 - CacheContentIndexTest and SimpleCacheSpanTest both work fine with Gradle
   but fail with seemingly valid test failures on Blaze.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=185366678

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=185139106

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=185138158

If the period uid doesn't match, the update procedure currently doesn't
remove the correct periods. This may cause the player to get stuck or to play
the wrong periods.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=185129503

Issue:#3816

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=185124378

Device models are from
https://support.google.com/googleplay/answer/1727131?hl=en. It looks like among
these devices Build.MANUFACTURER can be set to either "PHILIPS" or "Philips",
based on looking at internal bug reports.

Issue: #3807

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=185121182

We now build one .so file for the opus extension in the
internal build, so make the external build work the same way.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184962896

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184844870

Allow skipping an ad group when requested by IMA, even if we aren't currently
playing one, to handle cases where no ads in an ad group will load (so IMA
requests resuming content but we never managed to start playing an ad).

Use the known ad group index (rather than the expected one) when handling ad
group load errors. This ensures we skip the right ad group if we notify IMA of
playback errors for every ad in the ad group, then IMA notifies that the ad
group is empty via a load error.

Also make some other miscellaneous small fixes to ads code:

- Avoid warning about unexpected ad group indices more than once.
- Output a warning if the ad count in an ad group decreases.
- Remove unnecessary assertion.
- Fix getting the ad duration for ad indices that haven't loaded yet.
- Allow setting an ad group state to its current value.
- Fix javadoc for setting the ad resume position.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184831495

The ad index in the ad group may need to skip over ads that failed to load, so
it can't just be incremented any more.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184812556

When using cronet data source, calling read after the end of input has
been read will trigger this.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184667794

This feature is supported in the ConcatenatingMediaSource and is easily copied to this
media source. Also adding tests to check whether the atomic property works in normal
concatenation and in also in nested use.

Also fixes a bug where timeline methods of the DeferredTimeline were not correctly
forwarded.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184526881

This could happen when a media source is removed from a
DynamicConcatenatingMediaSource and one of its media periods is still active.
This media period is only removed by the player after the player received
a timeline update and thus we shouldn't release the removed child source
as long as it has active media periods.

Issue:#3796

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184522836

If IMA loads an empty VAST document for an ad group it notifies via a LOG
AdEvent. Handle the event by updating the AdPlaybackState accordingly.

The error state will be handled in ExoPlayerImplInternal in a separate change.

Issue: #3584

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184516585

Resolve the media period for ad playback when resolving a subsequent period and
when receiving a timeline where the playing period in range (but wasn't before).

Fix the seek position calculation when a current ad must be skipped and is
followed by another ad.

Check MediaPeriodInfos match when checking MediaPeriodHolders, to handle cases
where a future ad should no longer be played. This may involve playing two
content media periods consecutively.

Issue: #3584

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184514558

DefaultExtractorInput.SCRATCH_SPACE buffer is used to skip data by
reading it into this buffer and discarding.

Simultaneous use of skip methods corrupts this buffer. Normally the
read data is discarded so it doesn't matter but the underlying
DataSource may use the buffer too. If it's a CacheDataSource it uses
this buffer to read data from upstream then write to cache.

Issue: #3762

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184502170

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184501752

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184317120

This avoids reading a format that is not equal because of switching between
NO_VALUE and 0.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184298076

This makes assertion errors in code running on the Looper less easy to miss.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184294290

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184274139

This should be a no-op change.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184150266

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184125114

Even if a developer synchronizes every method, thread safety is still not guaranteed
because the internal callback methods can't be synced from outside.

Issue:#3773

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184122072

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184122057

1. Add string for STATE_CANCELED. Lint doesn't like that the
   switch statement on the state IntDef doesn't have a case
   for STATE_CANCELED. May as well add one, even if we're not
   planning on our demo app showing notifications for this
   state.
2. Replace non-human-readable error message with one provided
   by ErrorMessageProvider.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184120892

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184120873

In the case of the components we deliberately access via
reflection, it's normal that they might not be resolved
due to proguarding (i.e. if the app isn't being built to
include them). Don't note their omission.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184120611

The pending exception will be thrown upon returning to Java
from native, but we should return early rather than continuing
to execute the native method to the end so as to avoid undefined
behavior. Note that the return value is irrelevant (because the
pending exception will be thrown).

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184119606

Lint recommends switching to SparseArray<X> instead.
This is done for the DASH case. For the Cast case it's
easier to use a switch statement.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184119312

Note this case is special in that JUnit is an actual
compile dependency, as opposed to a testCompile or
androidTestCompile dependency.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184119235

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184119097

I think (?) they're harmless, but lint doesn't like them.
Using them within the class body means the TargetApi
annotation applies, which makes lint happy.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184117951

In MediaCodecRenderer, we currently uses codec.getInput/OutputBuffers event
though these APIs are deprecated and are not recommended from API 21+. This
change makes sure that:
- On API 20 and below, we will keep using codec.getInput/OutputBuffers.
- On API 21+, we will use getInput/OutputBuffer(index) APIs instead.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184112329

*** Reason for rollback ***

Broke everything

*** Original change description ***

Clean up message naming in EPII

***

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184061352

This applies when PlayerControlView is used as a standalone
component (not inside PlayerView). Previously hideAtMs was
set to 0, which caused the view to be immediatley hidden in
onAttachedToWindow. After this change the first time the view
is attached to the window is effectively treated as a "user
interaction" for the purposes of deciding when to timeout.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184056324

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184056034

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=184022988

This make sure all media sources can be reprepared after being released.

Issue:#3498

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=183990416

When the dynamic media source contains multiple empty timelines in a row and some
of them dynamically change to a non-empty timeline, the window and period indices
are not updated correctly because the index of the changed child source is wrong.

To fix this bug, the child index is added to the media period holder to have direct
access on the current child index to prevent ambiguity.

Furthermore, the uid is changed to be the hash code of the MediaSourceHolder not the
MediaSource itself to allow adding the same MediaSource twice without violating the
unique uid policy.

-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=183973462