Scheduler type annotations

[fjetter]

Since the introduction of the Task class we are keeping redundant dependeny sets on the Scheduler. I wanted to look into this since this is one of the major sources of memory use on the scheduler today. This would trigger a moderate refactoring that would replace the TaskState.dependency sets again with keys instead of the TaskState classes themselves. That would also be a change that would help a lot with GC duration.

Before I venture into this area I wanted to have mypy running

[github-actions]

Unit Test Results

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

    27 files  ± 0      27 suites  ±0   11h 36m 54s ⏱️ + 3m 24s
 4 105 tests  -  1   3 992 ✅ ± 0    112 💤 ±0  1 ❌  - 1 
51 465 runs   - 13  49 155 ✅  - 11  2 309 💤 ±0  1 ❌  - 2 

For more details on these failures, see this check.

Results for commit 2fd3de5. ± Comparison against base commit c5ca1ff.

This pull request removes 1 test.

distributed.tests.test_scheduler ‑ test_scheduler_close_fast_deprecated

♻️ This comment has been updated with latest results.