[FLINK-40671][network] Document why UnknownInputChannel graduation hardcodes needsRecovery=false

[1996fanrui]

What is the purpose of the change

UnknownInputChannel.toRemoteInputChannel / toLocalInputChannel hardcode
needsRecovery=false when graduating to a real channel. This looks unsafe next to the
recovery path, but it is correct: unknown channels only exist in BATCH jobs, which have no
channel state, while channel-state recovery only exists in streaming (unaligned checkpoints) —
the two are mutually exclusive. This adds a comment documenting that invariant.

Brief change log

• Comment-only, at the two graduation sites in UnknownInputChannel.

Verifying this change

This change is a trivial rework / code cleanup without any test coverage.

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

• Dependencies (does it add or upgrade a dependency): no
• The public API, i.e., is any changed class annotated with @Public(Evolving): no
• The serializers: no
• The runtime per-record code paths (performance sensitive): no
• Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn, ZooKeeper: no
• The S3 file system connector: no

Documentation

• Does this pull request introduce a new feature? no

[flinkbot]

CI report:

• 80ddbbb Azure: SUCCESS

Bot commands

The @flinkbot bot supports the following commands:

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

[davidradl]

I was expecting a change to the docs from the PR title. Is there an external consideration that we could add to the docs around this? Or is this just for code hygene?