[Feature] PIP-30: Introduce coordinator-based commit for Flink sink

[ifndef-SleePy]

Search before asking

• I searched in the issues and found nothing similar.

Motivation

This is the umbrella issue for PIP-30 — moving Paimon's Flink committer from a downstream CommitterOperator to a JobManager-side OperatorCoordinator. The target is Flink jobs writing to unaware-bucket append tables, where this enables region failover and removes the commit-side fan-in shuffle for large-parallelism Flink-to-Paimon write jobs.

Recent design discussion (with updated diagrams that didn't render cleanly on cwiki):

• Mailing-list thread: https://lists.apache.org/thread/5n1mwc7wc3cfjogz56w9k5ql3zyl00lw
• Updated design doc: https://docs.google.com/document/d/1asWKzoytfeB1D8bS_yRIAHnpR40frLf0fnPn2-WSL74/edit

Solution

The work will be split across a series of PRs to keep each one independently reviewable. Tentative breakdown (subject to change as reviews progress):

• Decouple Committer.Context from operator-only assumptions.
• Introduce WriteOperatorCoordinator as the JobManager-side commit driver.
• Wire the unaware-bucket append table to coordinator-based commit.
• Follow-ups (performance polish, documentation) as needed.

Links to per-PR issues will be added here as they are opened.

Anything else?

No response

Are you willing to submit a PR?

• I'm willing to submit a PR!

[ifndef-SleePy]

cc fishfishfishfishaa