No longer double count transfer cost in stealing

[hendrikmakait]

• Closes Stealing balance not accounting for recent decisions #7002
• Supersedes No longer double count transfer cost in stealing #7026

• Tests added / passed
• Passes pre-commit run --all-files

[github-actions]

Unit Test Results

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

       15 files  ±  0         15 suites  ±0   6h 11m 48s ⏱️ - 1m 25s
  3 107 tests +  6    3 017 ✔️ +  2    85 💤 ±0  5 ❌ +4 
22 997 runs  +43  22 086 ✔️ +39  906 💤 ±0  5 ❌ +4 

For more details on these failures, see this check.

Results for commit 9c9438e. ± Comparison against base commit 1fd07f0.

♻️ This comment has been updated with latest results.

[fjetter]

Codecov says this is not hit but I'd expect this to be hit by test_dont_steal_long_running_tasks

We should probably look at that test, maybe some test assumption is no longer true

[fjetter]

Ok, test_dont_steal_long_running_tasks is pretty meaningless. I believe something went wrong during a refactoring of that test. It only asserts that long_tasks are only "executed" once, i.e. they are not rescheduled. This is entirely irrelevant for stealing.

[fjetter]

Given that this test was already this way I suggest to set this out of scope for this PR. I might be able to address this with #7030

[fjetter]

All failing tests are known offenders except of test_wait_first_completed but I opened #7039 to make that one more robust

[hendrikmakait]

A/B test: https://github.com/coiled/coiled-runtime/actions/runs/3067074093