[zephyr] skip status log when no counters recorded

[wmoss]

Fixes #5175.

Map-only stages don't run their output through StatisticsGenerator, so the periodic status logger in subprocess_worker._periodic_status_logger kept emitting items=0 (0.0/s), bytes_processed=0.0MiB for the entire shard runtime — noisy and misleading. Now we skip the log line entirely until at least one of the item/byte counters has been recorded.

Generated with Claude Code

[claude]

Claude finished @wmoss's task in 2m 10s —— View job

---

Code review

No issues found. Checked for bugs and CLAUDE.md/AGENTS.md compliance.

The change is minimal and well-scoped:

• _periodic_status_logger short-circuits with continue when neither item_key nor byte_key has been recorded in ctx._counters. Once either appears, normal logging resumes — so map-only stages stay quiet without affecting stages that do flow through StatisticsGenerator.

• The two new tests cover both branches (skip on empty counters, log once at least one is present), use caplog filtered by message substring, and stop the daemon thread cleanly via stop_event.

• Read PR diff and context

• Read AGENTS.md files for changed paths

• Audit for CLAUDE.md/AGENTS.md compliance

• Scan for bugs/logic issues

• Validate any issues found

• Post review summary
  • Branch

[ravwojdyla]

We need to update the coordinator as well?