* Bump dataflow container version

* Container is 20210216

[BEAM-11811] [Dataflow Java] Don't allow numWorkers > maxNumWorkers.

* Renamed release/build.gradle to build.gradle.kts

* Migrated release module build file to kotlin script

Minor fixes for python nexmark benchmarks

* [BEAM-11747] Reject the mixed Java UDF and ZetaSQL builtin operators cases

* Move payload providers out into their own module.

This prevents all dependents on sdks:java:extensions:sql from needing to add a thrift and protobuf dependency.

* fix package

* restructure to put thrift and protobuf suppliers with their relevant packages

* fix moved messages

* fix moved messages

* fix moved messages

* Fix generated regex

* Add thrift suppresssion

* [BEAM-11806] Added explicit partition support for ProducerRecord instance with the KafkaIO.WriteRecords transform

* [BEAM-11806] Added custom partitioning support/tests for KafkaIO

[BEAM-11805] Revert version updates from #13765

1. Moved the cleanup mock for the flaky test down to the statement we
   want to check instead of placing an assert_not_called before the
   statement that we cannot guarantee in the testing environment.
2. The culprit statement could be the `set_cache_manager` when in the
   testing environment, the global singleton ie.current_env() holds a
   non empty cache manager for some reason while a new cache manager
   set would trigger a cleanup.
3. Moved some of the set_cache_manager tests to use local
   InteractiveEnvironment instance rather than the global one.

This reverts commit 5136c4ea.

[BEAM-11520] Stage extra PyPI dependencies with generated requirements

Merge pull request #13977 from [BEAM-11807] Fix issue with sdk worker multi-threading and boto3 client

[BEAM-11807] Fix issue with sdk worker multi-threading and boto3 client

apache_beam.io.gcp.spanner: Documentation update

This reverts commit e0ba4dc6.

[BEAM-10925] Don't publish udf-test-provider to Maven.

Merge pull request #13859 from [BEAM-11772] Integrate BigQuery sink file loads with GroupIntoBatches

[BEAM-11772] Integrate BigQuery sink file loads with GroupIntoBatches

[BEAM-11791] Continuously running microbenchmark for FnApiRunner

* Do not call `translations.optimize_pipeline` in Dataflow Streaming

* Lint