The skipped frame count (dropped frames due to encoding being
overloaded) would erroneously include lagged frames (dropped frames due
to render stalls).  This will make diagnosing actual issues a user might
be having a bit easier.

Problem:

When an output is started with encoders that have already been started
by another output, and it starts in between the window in between where
the first audio packets start coming in and where the first video packet
comes in, the output would get audio packets with timestamps potentially
much later than the first video frame when the first video frame finally
comes in.  The audio/video encoders will almost always have a differing
delay.

Solution:

Detect that starting window, and if within that starting window, wait
for a new keyframe from video instead of trying to sync up video.

Additional Notes:

In these cases where an output starts with already-active encoders, this
patch also reduces the potential sync offset between the first video
frame and the first audio frame.  Before, it would sync the first video
frame with the first audio frames right after that, but now it syncs
with the closest audio frame in the interleaved array, which halves the
potential sync difference between the first video frame and the first
audio frame of already-active encoders.  (So now the potential sync
difference is roughly 11.6 milliseconds at 44.1khz audio, where it was
23.2 before)

Ensures that the packet dts_usec vals which are generated for
syncing/interleaving use the proper offset relative to where they're
supposed to be starting from.  The negative DTS of a first video packet
could potentially have been applied twice due to this.

Fixes potential race conditions when two threads are trying to
initialize/start/stop the same encoders at the same time.

Allows the ability to use fixed stack memory to construct a calldata_t
structure rather than having to allocate each time.  This is fine to do
for certain signals where the calldata never goes above a specific size.
If by some chance the size is insufficient, it will output a log
message.

Allows getting the timestamp for the video frame currently being
rendered.

These functions allow animated gifs and management of image files.  Uses
libnsgif to load animated gifs.

Allows loading the file to memory first so the graphics subsystem
doesn't necessarily have to be locked when loading image files.

There shouldn't be any need to call these functions manually.  These
functions are called automatically by obs_save_sources and
obs_load_sources.

Originally this was programmed to call the recursive height/width
functions if the source type was an input with the intention of not
calling it on filters, but instead of doing that just program it to do
just that: only call the recursive height/width functions if it's not a
filter.

Allows transitions to be able to return a blank texture to transitions
if a transition is transitioning from/to a NULL source.

Allows gs_render_sprite to be used without a texture as long as cx/cy
are set.  There's no reason why it shouldn't be able to do this.

Needless function calls here for what should be intrinsic calls

These were in the APPLE section by mistake

Prevents from having to query the base audio subsystem for the sample
rate each time the function is called

Useful for going traversing the list of audio sources (particularly for
the new audio subsystem)

This was originally used for calculating audio volume if transitions
were active, but transitions won't work that way so tracking the active
transitions is no longer needed.

Uses 128bit integer math to prevent overflow

Allows some limited 128bit integer math functions, primarily for the
sake of performing calculations on nanosecond-based intervals of time

Before if a source was set to invisible it would still be considered
active.  This changes it so that the source is deactivated when the
source is invisible to reduce needless resource usage or capturing.

Renames:
----------------------------------------
obs_source_add_child
obs_source_remove_child
obs_source_enum_sources
obs_source_enum_tree
obs_source_info::enum_sources

To:
----------------------------------------
obs_source_add_active_child
obs_source_remove_active_child
obs_source_enum_active_sources
obs_source_enum_active_tree
obs_source_info::enum_active_sources

These functions/callbacks had misleading names: they originally implied
any child sources, when they actually meant active child sources that
are being used to render video or audio.  It's important that the
function names represent their actual purpose.

Prunes code used to workaround libobs "user sources" (such as
sourceSceneRefs) and instead simply holds its own references and saves
sources it chooses with obs_save_sources_filtered.

(Note: This commit breaks UI compilation.  Skip if bisecting)

Ensures that a reference is kept until the signals associated with that
reference are disconnected.

(Note: This commit breaks UI compilation.  Skip if bisecting)

Adds a means of saving specific sources that the front-end chooses,
rather than being forced to use the now-removed "user list".

(Note: This commit breaks UI compilation.  Skip if bisecting)

API Removed:
------------------------
obs_add_source

API Changed:
------------------------
obs_source_remove: Now just marks/signals a source for removal

The concept of "user sources" is flawed: it was something that the
front-end was forced to deal with if it wanted to automate source
saving/loading, and often it had to code around it.  That's not how
saving/loading should work, a front-end should be allowed to manage
lists of sources in the way it explicitly chooses, and it should be able
to choose which sources it wants to save/load.

These signals are meant to replace the add/remove signals.

These functions created stack variables but never actually initialized
them.  If the calldata variable is invalid, the return values will be
the uninitialized stack value.

The name was inconsistent with the darray_swap function that it's a
macro for.