Close #6329

In the example of 6329 an upstream failure generates a bunch of partially satisfied tasks - i.e. active (n=0) but not queued. Triggering them manually causes them to be added to a forced_released list - wrongly as they were not queued in the first place.

A bit later, the forced_released tasks top up the list of naturally released tasks. But by then there is some overlap between the two lists, so the queue limit matters (along with some task-state related race condition which I haven't fully understood, but that doesn't matter) as to how many of these tasks end up getting submitted again.

Check List

• I have read CONTRIBUTING.md and added my name as a Code Contributor.
• Contains logically grouped changes (else tidy your branch by rebase).
• Does not contain off-topic changes (use other PRs for other changes).
• Applied any dependency changes to both setup.cfg (and conda-environment.yml if present).
• Tests are included (or explain why tests are not needed).
• Changelog entry included if this is a change that can affect users
• Cylc-Doc pull request opened if required at cylc/cylc-doc/pull/XXXX.
• If this is a bug fix, PR should be raised against the relevant ?.?.x branch.