feat(sense): parallelize HandlerQueue per-handler concurrency
HandlerQueue now supports N concurrent slots per handler, configurable per
handler in journal.json via {handler}.max_concurrent. Default stays N=1
(byte-identical to prior behavior); operators raise it where API quota and
hardware allow. observe.status payload changes: handler_status["running"]
is now a list (length 0..N); top.py and apps/health/workspace.html updated
to render the count.
Co-Authored-By: Codex <codex@openai.com>