[FLINK-32828][1.18] Partition aware watermark not handled correctly shortly after job start up from checkpoint or savepoint

[pnowojski]

This is a backport of https://github.com/apache/flink/pull/24794 to 1.18

What is the purpose of the change

Properly initialize initial splits in WatermarkOutputMultiplexer

Without this fix, initial splits were registered in the multiplexer only
when first record from that split has been emitted. This was leading to
incorrectly emitted watermarks, as resulting watermark was not properly
combined from the initial splits, but only from the splits that have
already emitted at least one record.

Brief change log

Please check individual commit messages

Verifying this change

This bugfix is covered against future regressions by a new test.

Does this pull request potentially affect one of the following parts:

• Dependencies (does it add or upgrade a dependency): (yes / no)
• The public API, i.e., is any changed class annotated with @Public(Evolving): (yes / no)
• The serializers: (yes / no / don't know)
• The runtime per-record code paths (performance sensitive): (yes / no / don't know)
• Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn, ZooKeeper: (yes / no / don't know)
• The S3 file system connector: (yes / no / don't know)

Documentation

• Does this pull request introduce a new feature? (yes / no)
• If yes, how is the feature documented? (not applicable / docs / JavaDocs / not documented)

[flinkbot]

CI report:

• 1e368aa9d253f0fccd2a40148a5b80a2249cf49c Azure: SUCCESS

Bot commands

The @flinkbot bot supports the following commands:

• @flinkbot run azure re-run the last Azure build