Add metric for aborted compactions because of write load

## Summary

Very simple - whenever a compaction is aborted, increment a counter exposed in the `_stats` output.

## Additional context

Better visibility of this corner case via standard system monitoring tooling.

The issue names the compaction path and the `_stats` output but no files or tests. Locate where compactions are aborted because of write load, then trace how statistics are assembled. Done means the aborted-compaction counter appears in `_stats` and increments for that case.

Written by the indexing model from the issue text.