Add a criterion bench for the distributed cache hot path

[christian-schilling]

Add a criterion bench for the distributed cache hot path

deephistory_subdir_distributed mirrors deephistory_subdir_sparse -- the same long,
sparse-output history filtered by :/dir_00 -- but the cache stack under measurement carries a
DistributedCacheBackend in addition to sled, matching the josh CLI default.

The history is linear, so almost every input commit is ineligible for the sparse persisted
cache. The distributed backend then does nothing per commit except run its eligibility check,
which makes that check the dominant per-commit cost and the signal this bench isolates. Each
timed iteration resets the sled cache and deletes refs/josh/cache/* so the distributed cache
starts cold.

Change: bench-distributed-cache
Assisted-By: Claude Fable 5 (claude-fable-5)

[codspeed-hq]

Merging this PR will not alter performance

✅ 16 untouched benchmarks
🆕 3 new benchmarks

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
🆕	deephistory_subdir_distributed[100]	N/A	11.8 ms	N/A
🆕	deephistory_subdir_distributed[1000]	N/A	107.5 ms	N/A
🆕	deephistory_subdir_distributed[10000]	N/A	1.1 s	N/A

---

_{Comparing @changes/master/christian.schilling.de@gmail.com/bench-distributed-cache (9b71d13) with master (d7649b7)¹}

Footnotes

1. No successful run was found on master (dd692e2) during the generation of this report, so d7649b7 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩