docs: exploratory testing pass for persistence under a full disk (2026-10-10)

[jonbaldie]

Summary

Adds the 2026-10-10 exploratory testing report, its driver scripts and transcripts, and an index entry.

The pass covered three journeys: polling workers on the compiled binary, --persist when the PERSIST volume fills up, and the README's Docker volume deployment. It confirmed two data-loss bugs, both filed:

• Short writes on a full PERSIST volume are reported as success and corrupt persist.dat, losing acknowledged items #160: short writes on a full volume are reported as success and corrupt persist.dat. Items acknowledged with 200 are lost after a crash, or by a graceful shutdown that exits 0.
• Enqueue/dequeue return 500 when the persist write fails but have already changed the queue (rejected items delivered, dequeued items lost) #161: when the persist write fails, enqueue and dequeue return 500 but the in-memory queue has already changed.

This PR is documentation only. It does not fix either issue.

Test plan

• Docs-only change. CI must pass.

🤖 Generated with Claude Code