fix(logger): bound flush() wait with timeout to prevent caller deadlock (#187)

[SAY-5]

Summary

Fixes #187 — flush() could block its caller indefinitely if the output worker thread became wedged (deadlocked, sigstop'd, or blocked on a slow stdout pipe). The previous ack_rx.recv() was an unbounded wait, so any worker stall would deadlock process shutdown.

Change

• Add FLUSH_TIMEOUT = 5s constant.
• Replace ack_rx.recv() with ack_rx.recv_timeout(FLUSH_TIMEOUT).

A healthy worker draining typical bursts of buffered writes finishes well inside the budget. A genuinely wedged worker now loses at most a small tail of output instead of hanging the caller forever — generally the right tradeoff for shutdown paths.

Test plan

• cargo build --lib — clean
• cargo test --lib infra::logger -- --test-threads=1 — 16/16 pass
• test_flush_completes_without_panic still passes
• test_thread_safety (which calls flush() after concurrent writers) still passes

Note: test_create_logger_uses_global_level is flaky under parallel cargo test because several logger tests mutate the set_global_level static; unrelated to this change and reproducible on main.

[TrueNine]

Merged into dev via cherry-pick. The PR base was changed from main to dev, which caused conflicts because dev had diverged significantly. All commits have been cherry-picked onto dev and pushed. Thanks for the contributions!

[TrueNine]

Thank you for the contribution! All commits have been cherry-picked and merged into the dev branch. 🙏