fix: make AIOKafkaProducer.start() idempotent to prevent busy loop

[gloryfromca]

When start() is called multiple times on the same producer instance (e.g., by multiple listeners sharing the same producer), each call would create a new sender task. Multiple sender tasks sharing the same message accumulator causes them to wake each other up via drain_by_nodes(), resulting in an infinite busy loop.

Changes:

• Add _started flag to track producer startup state
• Return early if start() is called on an already-started producer
• Reset _started flag on startup failure to allow retry

This ensures only one sender task is created per producer instance.

Changes

Fixes #

Checklist

• I think the code is well written
• Unit tests for the changes exist
• Documentation reflects the changes
• Add a new news fragment into the CHANGES folder
  • name it <issue_id>.<type> (e.g. 588.bugfix)
  • if you don't have an issue_id change it to the pr id after creating the PR
  • ensure type is one of the following:
    • .feature: Signifying a new feature.
    • .bugfix: Signifying a bug fix.
    • .doc: Signifying a documentation improvement.
    • .removal: Signifying a deprecation or removal of public API.
    • .misc: A ticket has been closed, but it is not of interest to users.
  • Make sure to use full sentences with correct case and punctuation, for example: Fix issue with non-ascii contents in doctest text files.