Add a metric for CPU time per thread type

`synapse_thread_cpu_seconds_total{thread_type, mode}` reads
`/proc/self/task` on each scrape and groups threads into the main thread,
each registered threadpool, tokio (by kernel thread name), and `other`.

The kernel only reports CPU time for live threads, so the collector sums
per-thread deltas between scrapes. A thread that exits loses the CPU it
used since the last scrape.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EDruzqbGMzEHotouLuudxP

scripts-dev/mypy_synapse_plugin.py

@@ -147,6 +147,7 @@ prometheus_metric_fullname_to_label_arg_map: Mapping[str, ArgLocation | None] =
     "synapse.metrics._gc.PyPyGCStats": None,
     "synapse.metrics._reactor_metrics.ReactorLastSeenMetric": None,
     "synapse.metrics.CPUMetrics": None,
+    "synapse.metrics._thread_cpu.ThreadCPUMetrics": None,
     "synapse.metrics.jemalloc.JemallocCollector": None,
     "synapse.util.metrics.DynamicCollectorRegistry": None,
     "synapse.metrics.background_process_metrics._Collector": None,