Kafka Streams runner: expose topic-level configuration for runner-created topics

## Summary

The Kafka Streams runner creates its own internal topics (a bootstrap topic per Impulse and Read, a repartition topic per GroupByKey). Today only the partition count (`--internalParallelism`) and replication factor are configurable.

A user should be able to set the rest of the topic-level configuration — retention, compaction, segment settings, min.insync.replicas and so on. Two options discussed on https://github.com/apache/beam/pull/39546: pass a JSON-serialized dictionary of topic configs, and/or expose the most common ones as separate pipeline options.

Part of #18479.

Start with the Kafka Streams runner and its handling of runner-created bootstrap and repartition topics for Impulse, Read, and GroupByKey. Review the existing --internalParallelism and replication-factor pipeline options, then compare the JSON topic-config and separate-option approaches discussed in PR 39546. Done means supported topic-level settings are exposed and applied to those internal topics, with tests covering the configuration.

Written by the indexing model from the issue text.