Replacing the EventCollector and the MemoryArchivist.

Finished EventCollector and adjusted ExecutionGraph and ExecutionVertex to register actors as listeners.

This commit is based on a diff by @fpompermaier (see https://github.com/apache/incubator-flink/pull/220).

This closes #220

This closes #232

- update _config.yaml of integrated documentation
- quick check on the quickstarts / build instructions to match the latest release
- quickstart.sh scrips now point to the correct versions.

This is a combination of commits:
* make hadoop-compat compatible to hadoop yarn and remove org.apache.hadoop code
* Extended usercode wrapper, fixed IF serialization
* introduce pluggable type converter
* make the converter interface more generic; (hopefully) improved fetching logic

Added 'AvroInputFormat' with test.

Minor cleanups.

This is a combination of 5 commits:
extended submission client

minor yarn refinements

some minor fixes

moved print usage

based on https://github.com/hortonworks/simple-yarn-app

Build debian package only with special profile for that (to speed up the build)
Build uber-jar of stratosphere (for super-handy yarn deployment)

Old commit messages (from rebase:)

advanced configuration file handling (maven and jar magic)
Command line arguments for client (not yet done)

Change JobManager startup to allow external control
Use environment variables to pass variables to ApplicationMaster
Use contstants for settings
Fix CLI options
minor refinemens (pact-hbase dependencies)

Yarn runs with HDFS
jar files and configs are now shared using hdfs
many many deployment bug fixes
uberjar is automatically uploaded to dopa server

It contains code maintained by the project. (Most of them have a lot of dependencies and are not required by most users)
Move pact-array-datamodel to stratosphere-addons
Move pact-hbase to stratosphere-addons

End the version name craziness and go back to have 0.4-SNAPSHOT as
version number. Also change the name of the toplevel pom back to
stratosphere.

1) Changed profile activation to properties (like hbase) 2) moved all properties to top of pom 3) moved module list to top of pom

Added extra profile entries in the pom.xml files in order to fix the hadoop_v1 compilation problems.