parsec: catch section/option conflict in validation

[oliver-sanders]

• Closes Parsec: accidentally specifying sections as settings not handled properly #4955
• If an item in the spec is a section but the user specified an option, raise an IllegalItemError if it is.

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).
• CHANGES.md 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]

I think it's worth considering whether the integration test is ok.

[wxtim]

I'm happy with this -
I think it's big enough to be worth tagging @MetRonnie or @hjoliver (Who-ever gets there first) as second review.

[MetRonnie]

Tested out a few examples.

I think some items can be tripped up earlier by the deprecation upgrader. E.g.

# flow.cylc
scheduling = 22

  File "~/cylc-flow-8.1/cylc/flow/scripts/validate.py", line 154, in wrapped_main
    cfg = WorkflowConfig(
  File "~/cylc-flow-8.1/cylc/flow/config.py", line 308, in __init__
    self.pcfg = RawWorkflowConfig(
  File "~/cylc-flow-8.1/cylc/flow/cfgspec/workflow.py", line 2042, in __init__
    self.loadcfg(fpath, "workflow definition")
  File "~/cylc-flow-8.1/cylc/flow/parsec/config.py", line 84, in loadcfg
    self.upgrader(sparse, title)
  File "~/cylc-flow-8.1/cylc/flow/cfgspec/workflow.py", line 1891, in upg
    u.upgrade()
  File "~/cylc-flow-8.1/cylc/flow/parsec/upgrade.py", line 200, in upgrade
    old = self.get_item(upg['old'])
  File "~/cylc-flow-8.1/cylc/flow/parsec/upgrade.py", line 107, in get_item
    item = item[key]
TypeError: string indices must be integers

Also I noticed the opposite (using a setting as a section) gives traceback too, seems like a low priority though

# flow.cylc
[scheduling]
    [[cycling mode]]

  File "~/cylc-flow-8.1/cylc/flow/scripts/validate.py", line 154, in wrapped_main
    cfg = WorkflowConfig(
  File "~/cylc-flow-8.1/cylc/flow/config.py", line 308, in __init__
    self.pcfg = RawWorkflowConfig(
  File "~/cylc-flow-8.1/cylc/flow/cfgspec/workflow.py", line 2042, in __init__
    self.loadcfg(fpath, "workflow definition")
  File "~/cylc-flow-8.1/cylc/flow/parsec/config.py", line 86, in loadcfg
    self.validate(sparse)
  File "~/cylc-flow-8.1/cylc/flow/parsec/config.py", line 96, in validate
    return self.validator(sparse, self.spec)
  File "~/cylc-flow-8.1/cylc/flow/parsec/validate.py", line 1130, in cylc_config_validate
    return CylcConfigValidator().validate(cfg_root, spec_root)
  File "~/cylc-flow-8.1/cylc/flow/parsec/validate.py", line 216, in validate
    cfg[key] = self.coercers[specval.vdr](value, keys + [key])
  File "~/cylc-flow-8.1/cylc/flow/parsec/validate.py", line 389, in coerce_str
    value = cls.strip_and_unquote(keys, value)
  File "~/cylc-flow-8.1/cylc/flow/parsec/validate.py", line 528, in strip_and_unquote
    if value.startswith(substr):
AttributeError: 'OrderedDictWithDefaults' object has no attribute 'startswith'

[oliver-sanders]

Will haver a quick look to see what can be done for those cases.

[oliver-sanders]

@MetRonnie, I've expanded this PR to cover the two additional cases you found.

[MetRonnie]

Can't find a way to produce traceback anymore. Worst thing I could find was

# flow.cylc
[scheduling]
    [[max active cycle points]]

$ cylc validate Mist
WARNING - deprecated items were automatically upgraded in "workflow definition"
WARNING -  * (8.0.0) [scheduling]max active cycle points -> [scheduling]runahead limit - "n" -> "Pn"
WorkflowConfigError: bad runahead limit "POrderedDictWithDefaults()" for integer cycling type

But I think this is very niche and not worth addressing

[wxtim]

Just deconflicted CHANGES.md

Will merge once tests OK

One test failure - being dealt with elsewhere and unrelated to this pr