This adds dedicated WindowAssigners for processing time and event time.
timeWindow() and timeWindowAll() respect the TimeCharacteristic set
on the StreamExecutionEnvironment.

This will make the easy stuff easy, i.e. using time windows and quickly
switching the time characteristic. Users will then have the flexibility
to mix different kinds of window assigners in their job.

This also expands the translation tests to verify that the correct
window operators are instantiated.

[FLINK-3420] [api-breaking] Remove utility functions 'readTextFileWithValue' and 'readFileOfPrimitives' from StreamExecutionEnvironment

These methods are highly specific for very niche cases of bounded data stream processing.
As such, the disadvantages (bloat and lock the API, lock the development into support) outweigh the benefit.

This closes #1648

Because the implicit conversion creates a new ExecutionEnvironment, it leads to
strange errors when used withing programs with more than one source.

This closes #1563

This closes #1524

This closes #1635

Update index.js and reset vendor.css and vendor.js to master version

Update web-dashboard

Remove duplicate flink-runtime-web dependency from flink-tests

Remove not used ProgramStopException

Change stopping behaviour to only work in job status RUNNING

This closes #750.

- added JobType to JobGraph and ExecutionGraph
- added interface Stoppable, applied to SourceStreamTask
- added STOP signal logic to JobManager, TaskManager, ExecutionGraph
- extended Client to support stop
- extended Cli frontend, JobManager frontend
- updated documenation

Fix JobManagerTest.testStopSignal and testStopSignalFail

The StoppableInvokable could not be instantiated by Task because it was declared as a private
class. Adds additional checks to verify that the stop signal behaves correctly.

Auto-detect if job is stoppable

A job is stoppable iff all sources are stoppable

- Replace JobType by stoppable flag
- Add StoppableFunction and StoppableInvokable to support the optional stop operation
- added REST get/delete test (no extra YARN test -- think not required as get/delete is both tested)
- bug fix: job got canceld instead of stopped in web interface
- Add StoppingException
- Allow to stop jobs when they are not in state RUNNING

Second round of Till's comments

A RestartStrategy defines how the ExecutionGraph reacts in case of a restart. Different strategies
are conceivable. For example, no restart, fixed delay restart, exponential backoff restart, scaling
in/out restart, etc.

Expose RestartStrategy to user API

This removes the setNumberExecutionRetries and the setDelayBetweenRetries on the ExecutionEnvironment and
the ExecutionConfig. Instead the more general RestartStrategy can be set. In order to maintain the
separation between the runtime and api module, one sets a RestartStrategyConfiguration which is transformed
into a RestartStrategy on the JobManager.

Replace old execution-retries configuration parameters by restart-strategy.

Add FixedDelayRestartStrategy test case

Reintroduce old configuration values and API calls for the deprecated restart mechanism

The old configuration values and API calls will be respected if no explicit
RestartStrategy has been set. The values, if correct, are used to instantiate
a FixedDelayRestartStrategy.

Add deprecation comments to the JavaDocs

Add logging statement for job recovery

Fix JobManagerProcessFailureBatchRecoveryITCase by introducing a job recovery timeout

Add proper annotations to RestartStrategies

Let ExecutionGraphRestartTest extend TestLogger

This closes #1470.

Closes #1638

Adds a new per-job tab displaying subtask statistics aggregated by TaskManager.

This closes #1564.

This reverts commit cf3ae88b.

The reason is that we cannot be sure that a failing sender will always be the root
cause why the receiver cannot read the data. A TM can also fail after a sender is
finished leading to a loss of the intermediate result partitions. Then the receiver
will no longer be able to read the data.

[FLINK-3367] Add PublicEvolving and Internal annotations to flink-streaming-java and flink-streaming-scala

This closes #1606

This also moves the utilities (HDFSCopyFromLocal and HDFSCopyToLocal) to
the RocksDB package because we would need a HDFS dependency in
flink-core otherwise.

This enables efficient incremental aggregation of fold window.

This also adds:
- WindowedStream.apply(initVal, foldFunction, windowFunction)
- AllWindowedStream.apply(initVal, foldFunction, windowFunction)

This closes #1605

This closes #1533

This adds constructors that only take a backup dir URI and use it to initialize
both the RocksDB file backups and the FileSystem state backend for non-partitioned
state.

Also, the RocksDBStateBackend now automatically picks up the TaskManager's temp directories,
if no local storage directories are explicitly configured.