This closes #12907

This closes #12471

This close #12150

This closes #10536

 Brief change log:
  - remove the scala version suffix for connector-hive and queryable-state-client-java
  - add the scala dependencies for table-api-scala and flink-sql-connectors
  - correct the scala-free check logic in `verify_scala_suffixes.sh`

This commit splits the flink-table module into multiple submodules in
accordance with FLIP-32 (step 1).

The new module structure looks as follows:

flink-table-common
       ^
       |
flink-table-api-java <------- flink-table-api-scala
       ^                                 ^
       |                                 |
flink-table-api-java-bridge    flink-table-api-scala-bridge
       ^
       |
flink-table-planner

The module structure assumes that the type system has been reworked such
that only one table environment exists for both Java and Scala users.

The module `flink-table-planner` contains the content of the old
`flink-table` module. From there we can distribute ported classes to
their final module without breaking backwards compatibility or
force users to update their dependencies again.

For example, if a user wants to implement a pure table program in Scala,
`flink-table-api-scala` and `flink-table-planner` need to be
added to the project.

Until we support pure table programs, `flink-table-api-scala/java-bridge`
and `flink-table-planner` need to be added to the project.

This closes #7587.

This commit removes all classes and methods that have been deprecated in
Flink 1.6 for separating Kafka connectors from Avro and JSON formats.

This closes #7182.

In order to satisfy dependency convergence we need to exclude the kafka-clients from the base
flink-connector-kafka-x dependency in every flink-connector-kafka-y module.

This closes #7140.

This closes #6366.

This closes #5840.

This closes #5673.

This closes #5564.

This closes #5491.

By not setting Avro as 'provided', the build system will put it
into the user code fat jar, rather than assuming it will be part
of Flink's 'lib' folder.

That way Avro is loaded child-first through the user code class
loader, giving it independent separate copies per load that avoid
version conflicts and caching problems.

This removes all dependencies on Scala-dependent projects.

This commit introduces a hard wired test dependency to
'flink-test-utils_2.11' to avoid introducing a Scala version dependency
due to a non-exported test utility.

This also adds a new test that verifies that we correctly register
Avro Serializers when they are present and modifies an existing test to
verify that we correctly register dummy classes.

This might include some bugfixes

We use custom serializers to ensure that we have control over the
serialization format, which allows us easier evolution of the format in
the future.

This also implements custom serializers for KafkaProducer11, the only
TwoPhaseCommitSinkFunction we currently have.

Sometimes 1000m was not enough memory to run at-least-once tests with broker failures on Travis

This closes #4456.

This closes #4321