Before this, a sliding time window would keep emitting the last result
because the number of elements per pre-aggregation result was not
correctly reset on eviction.

Problem: Failures in the network stack were not properly handled and correctly
attributed.

Solution: Failures are always attributeed to the client (consumer). This change
introduces TransportException, which indicates whether the problem ocurred
locally or remotely. This makes it easy to reason about the source of a problem.

This closes #713.

Closes #697

This was using != before, now uses !(a.equals(b))

Problem: cancelling of tasks sometimes leads to misleading error messages about
"not found partitions". This is an artifact of task cancelling. If a task
(consumer) consumes data from another remote task (producer), its sends a
partition request over the network. If the producer fails concurrently with this
request, the request returns with a PartitioNotFoundException to the consumer.
If this error message is received *before* the consumer is cancelled (as a
result of the failing producer), you see the misleading error being attributed
to the consumer. This makes it hard to trace the root cause of the problem (the
failing producer).

Solution: when a consumer receives a remote PartitionNotFoundException, it asks
the central job manager whether the producer is still running or has failed.

If the producer is still running, the partition request is send again (using an
exponential back off). If the following requests fail again, the consumer fails
with a PartitionNotFoundException.

If the producer has failed, the consumer is cancelled.

If the producer is not running and has not failed, there is a bug either in the
consumer task setup (e.g. requesting a non-existing result) or in the network
stack (e.g. unsafe publication of produced results), in which case the error is
attributed to the consumer.

---

The new Akka messages introduced with this change are only exchanged in error
cases and don't affect normal operation.

Normal operation (not affected by this change):
	- TM1=>TM2: request result
	- TM2=>TM1: result

Error case:
	- TM1=>TM2: request result
	- TM2=>TM1: PartitionNotFoundException
	- TM1=>JM: check partition state
	- JM=>TM1: retrigger request -OR- cancel consumer

This closes #705.

Adds pipeline comments

Adds pipeline IT case

[FLINK-2050] [ml] Introduces new pipelining mechanism using implicit classes to wrap the algorithm logic

This closes #704.

Also added attribution for some of the Latex in optimization framework.

This closes #688.

Added test case for convergence

This closes #692.

This closes #698

This closes #682

 - Exclude static fields in Scala Pojo analysis
 - Recognize Java Tuples
 - Clean up legacy code

(And also make one of the field type retrieval methods nicer)

This closes #669

Improve type safety.
Minor cleanups in ProjectOperator.

[FLINK-1418] [apis] Fix eager print() and adjust all tests and examples to not fail due to "eager" print method

 - Add lastJobExecutionResult for getting the result of the last execution, when executing "eager" execution methods

This closes #699

print() now uses collect() internally

This still evaluates the old parameter, if the old one is set and the new one is not set.

This closes #703

The counts are off by 1 in some cases, so the test is not activated.
I commit it to allow others to use it as a base of investigation.

[FLINK-2057] [FLINK-2058] [core] Fix hadoop input split class loading and remove IOReadableWritable from InputSplits

* add a decorateInputStream() method as hook in FileInputFormat
* provide a InputStreamFSInputWrapper to conveniently wrap InputStreams
* base existing .deflate file support on these changes
* add a test to verify the decoration

This closes #658

This closes #701

[runtime] Improve error message for failed network connections due to limited number of file handles

This closes #686.

The following values are parsed as booleans:

"true" or "1" -> true
"false" or "0" -> false

All checks are performed case-insensitive.

This closes #685.

This closes #680

    deleted VertexWithValue type;
    deleted InaccessibleMethodException; if the options are not set, -1 is returned;
    added missing javadocs;
    added tests;
    renamed type parameters VertexKey -> K, VertexValue -> VV, EdgeValue -> EV.

This commit squashes the following:

Pratially addressed inline comments

Added test for removal of a non-SP-edge

Removed trailing splits into vertex key and value

Made the extensions optional and separate

Removed overhead when the options are not used

Added Algorithm end string to the library methods

This closes #625

Closes #676