fix(cdc): restart snapshot read correctly

[Copilot]

Cherry-pick of #26105 onto release-3.0. Fixes incorrect CDC backfill snapshot restart behavior when needs_rebuild_snapshot is set.

What's changed and what's your intention?

cdc_backfill.rs

• Move should_bypass_snapshot_stream_patch (rate-limit-to-zero check) inside backfill_loop so it is re-evaluated on each restart, not once at startup.
• Capture old rate_limit_rps state before updating to new value — the bypass flag must reflect whether the previous snapshot stream was ever established.
• Change barrier staging condition: if can_start_new_snapshot → if can_start_new_snapshot || needs_rebuild_snapshot, so a requested rebuild actually stages the barrier and breaks out to restart the snapshot.
• Remove the now-redundant if needs_rebuild_snapshot { continue 'backfill_loop; } inside the barrier emit path (was dead code after the condition fix above).
• Fix snapshot stream patch: resume the valve before consuming when the snapshot is paused, and add a drain arm for the paused case to avoid dropping the resumed chunk.

postgres_schema_check_and_schema_change.slt

• Rename mood_enum → mood_enum_2 (with DROP TYPE IF EXISTS guard) to avoid type name conflicts across parallel test runs.

Checklist

• I have written necessary rustdoc comments.
• I have added necessary unit tests and integration tests.
• I have added test labels as necessary.
• I have added fuzzing tests or opened an issue to track them.
• My PR contains breaking changes.
• My PR changes performance-critical code, so I will run (micro) benchmarks and present the results.
• I have checked the Release Timeline and Currently Supported Versions to determine which release branches I need to cherry-pick this PR into.

Documentation

• My PR needs documentation updates.

Release note

Fixed a bug where CDC table backfill would not correctly restart the snapshot read after a rate-limit change or explicit rebuild request, potentially causing data loss or stalled backfill progress during recovery.