Validation of order of initial/start/stop/final cycle points

[wxtim]

Closes #6866

Follow-up to #4302

Apparently there was no systematic validation of whether users were supplying sensible values for these, except in a small number of cases.

In this PR, I write a thorough set of cases and ensure that they all pass.

Currently this targets 8.4.x, but I understand that given the refactoring it might be sensible enough to target 8.6.0.

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.

[wxtim]

Broken the one test which did test any case

[MetRonnie]

I think there is conflation of 2 different cases here.

1. Valid but useless X cycle point. E.g.:

   start cycle point '1' before the initial cycle point '2'

   This should give a warning.

2. Invalid X cycle point. E.g.:

   stop cycle point '1' before the initial cycle point '2'

   I think this ought to raise an input error.

[wxtim]

There are already errors raised elsewhere for seriously bad cases (mostly final < initial).

I can't see, however why you think these two cases should be treated differently - Neither make a lick of sense, but I don't think one is more clearly wrong than the other?

[MetRonnie]

Case 2 is a user error that will result in the workflow shutting down immediately. It would be more helpful to flag up this error instead of just warning?

[wxtim]

will result in the workflow shutting down immediately

I think that's less dangerous than case 1 where the workflow runs, but if the user intended startcp=10 it won't work as expected, but will do so quietly.

[MetRonnie]

I think I am happy for case 1 to raise an error too. However we may want to discuss this

[oliver-sanders]

(assigning commenters as reviewers)

[wxtim]

Draft because it's broken loads of tests