[KafkaIO] Fix per-split metric updates for KafkaUnboundedReader and ReadFromKafkaDoFn

[sjvanrossum]

This reverts #31137 and #31281. The changes in the former PR overwrite the per-split metric backlog_bytes.${SPLIT} with the per-partition metric rather than the accumulated value for the split. The latter PR introduced a map to store metrics for all past and current splits (1 partition) of the ReadFromKafkaDoFn instance and may repeatedly overwrite non-current splits with stale values. The map used to store these values is not thread-safe and may trigger a ConcurrentModificationException since GetSize and other SDF methods may concurrently attempt to read and write the map.

Finally, the per-split caches kept by the instance are keyed on TopicPartition, which is not unique among all splits since the split may override the bootstrap server. This has been fixed for this cache, all other caches will be patched in a separate PR.

---

Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:

• Mention the appropriate issue in your description (for example: addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, comment fixes #<ISSUE NUMBER> instead.
• Update CHANGES.md with noteworthy changes.
• If this contribution is large, please file an Apache Individual Contributor License Agreement.

See the Contributor Guide for more tips on how to make review process smoother.

To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md

GitHub Actions Tests Status (on master branch)

See CI.md for more information about GitHub Actions CI or the workflows README to see a list of phrases to trigger workflows.

[github-actions]

Checks are failing. Will not request review until checks are succeeding. If you'd like to override that behavior, comment assign set of reviewers