Review user-defined fine performance metrics

[crusaderky]

User-defined metrics can

• use any msgpack-serializable Hashable as label
• use any arbitrary string as unit

This creates edge cases that insofar were poorly tested.

This PR also removes special handling of tuple labels, which caused the execute tuples to become of unpredictable length which created even more edge cases to be tested - this time, needlessly so.

[milesgranger]

Nice!

[github-actions]

Unit Test Results

See test report for an extended history of previous test failures. This is useful for diagnosing flaky tests.

       20 files  ±  0         20 suites  ±0   11h 42m 17s ⏱️ - 8m 52s
  3 677 tests +  2    3 567 ✔️ +  4     108 💤 ±0  2 ❌  - 2 
35 560 runs  +20  33 794 ✔️ +22  1 764 💤 ±0  2 ❌  - 2 

For more details on these failures, see this check.

Results for commit 4a1e50f. ± Comparison against base commit e31c864.

This pull request removes 1 and adds 3 tests. Note that renamed tests count towards both.

distributed.tests.test_scheduler ‑ test_cumulative_worker_metrics

distributed.tests.test_worker_metrics ‑ test_reschedule
distributed.tests.test_worker_metrics ‑ test_send_metrics_to_scheduler
distributed.tests.test_worker_metrics ‑ test_user_metrics_weird