[None][fix] Handle skipped stale SWA blocks during cleanup

[jiaganc]

@coderabbitai summary

Description

Decoder-side disaggregated requests disable SWA scratch reuse before receiving remote KV data. Their stale out-of-window blocks can still have no allocated page, but cleanup assumed every missing page represented a scratch block and asserted that scratch reuse was enabled.

Treat missing stale pages as already released in both commit cleanup and stale-block unlocking. Existing type and cache sanity checks continue to validate allocated pages and the overall cache state.

Test Coverage

• Ran the standalone real KV Cache Manager v2 reproducer on B200; it previously failed in _on_stop_committing() and now exits successfully.
• python3 -m py_compile tensorrt_llm/runtime/kv_cache_manager_v2/_core/_kv_cache.py
• Repository pre-commit hooks
• git diff --check

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• If PR introduces API changes, an appropriate PR label is added - either api-compatible or api-breaking. For api-breaking, include BREAKING in the PR title.

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• Update tava architecture diagram if there is a significant design change in PR.

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

To see a list of available CI bot commands, please comment /bot help.

[lfr-0531]

Closing this PR since these changes are included in #15710