This closes #2460.

The opt directory should be reserved for Flink JARs which users may
optionally move to lib to be loaded by the runtime. flink-gelly-examples
is a user program so is moved to the examples folder.

This closes #3691

If "<flink-dir>/build-target" already existed, running 'mvn package' for
flink-dist would create a symbolic link inside "<flink-dir>/build-target"
instead of replacing that symlink. This commit fixes this behaviour of 'ln -sf'
by adding the --no-dereference parameter.

This closes #3331

This also removes gelly-examples from 'opt' assembly

Package the Flink connectors, metrics, and libraries into subdirectories
of a new opt directory in the release/snapshot tarballs.

This closes #3014

Currently, every project in Flink has a hard (compile scope) dependency on the jsr305, slf4j, and log4j
artifacts. That way they are pulled into every fat jar, including user fat jars as soon as they refer to
a connector or library.

This commit changes the behavior in two ways:

  1. It removes the concrete logger dependencies from the root pom file. Instead, it adds them to the
     'flink-core' project. That way, all modules that refer to 'flink-core' will have those dependencies
     as well, but the projects that have 'flink-core' as provided (connectors, libraries, user programs,
     etc) will have those dependencies transitively as provided as well.

  2. The commit overrides the slf4j and jsr305 dependencies in the parents of 'flink-connectors',
     'flink-libraries', and 'flink-metrics' and sets the to 'provided'. That way all core projects
     pull the logger classes, but all projects that are not part of flink-dist (and rather bundled
     in fat jars) will not bundle these dependencies again.

The flink-dist puts the dependencies into the fat jar (slf4j, jsr305) or the lib folder (log4j).

This closes #2850

This also now packages the RocksDB state backend into the Flink distribution

This closes #2482

This adds back the link 'build-target' which points to the final build
assembly in 'flink-dist/target/'.

This closes #2620

On Windows, the plugin downloads and executes code from the author's
web site. The downloaded file is not signed in the same way as Maven
artifacts from Maven central which have to be signed with the
developer's key. This could be a potential target for attackers.

This closes #2586

Implemented Mesos AppMaster including:
- runners for AppMaster and TaskManager
- MesosFlinkResourceManager as a Mesos framework
- ZK persistent storage for Mesos tasks
- reusable scheduler actors for:
  - offer handling using Netflix Fenzo (LaunchCoordinator)
  - reconciliation (ReconciliationCoordinator)
  - task monitoring (TaskMonitor)
  - connection monitoring (ConnectionMonitor)
- lightweight HTTP server to serve artifacts to the Mesos fetcher (ArtifactServer)
- scenario-based logging for:
  - connectivity issues
  - offer handling (receive, process, decline, rescind, accept)
- incorporated FLINK-4152, FLINK-3904, FLINK-4141, FLINK-3675, FLINK-4166

This closes #2309

This closes #2324

- moved user-facing API to 'flink-metrics/flink-metrics-core'
- moved JMXReporter to 'flink-metrics/flink-metrics-jmx'
- moved remaining metric classes to 'flink-runtime'

This closes #2226

This closes #2042

This ensures that all properties of the root pom are properly
resolved by running the Shade plugin. Thus, our root pom does not have
to depend on a Scala version just because it holds the Scala version
properties.

This closes #1755

- add suffixes
- adapt change-scala-version script

This closes #1529.

[FLINK-3195] [examples] Consolidate batch examples into one project, unify batch and streaming examples under on parent project

This closes #1409

0.10-SNAPSHOT continues on branch release-0.10

This closes #1259

Update Scala 2.11 version and jline dependency.

This closes #1197

This closes #1194

This closes #1113

-----
Closing unresponsive PR:
This closes #862

This is needed because the Hadoop IF/OF's are using Hadoop's FileSystem stack, which is using
the security credentials passed in the JobConf / Job class in the getSplits() method.

Note that access to secured Hadoop 1.x using Hadoop IF/OF's is not possible with this change.
This limitation is due to missing methods in the old APIs.

- Add some comments & change dependency scope to test

 - Adds a separate Maven project for easier maintenance. Also allows users to refer to runtime without web libraries.
 - Simple HTTP server based on Netty HTTP (slim dependency, since we use netty anyways)
 - REST URL parsing via Netty Router
 - Abstract stubs for handlers that deal with errors and request/response
 - First set of URL request handlers that produce JSON responses

This closes #677
This closes #623
This closes #297

Closes #851

This closes #837.

This closes #783

This closes #691

This can either be used to connect to a Flink cluster or in local mode
with a cluster in the same JVM.

This only works for Scala 2.10 for now because the Scala REPL was
modified in 2.11.