fix: warn about clashing linter names

[hippietrail]

Issues

Resolves #2343

Description

In fact this finds 3 pairs of clashing linters already!

• Intact - one in ClosedCompounds, the other in phrase_corrections.
• Nobody - one in ClosedCompounds, the other standalone.
• OfCourse - one standalone, the other in phrase_corrections.

I have intentionally not tried to resolve any of these as part of this PR.

I included another warning in this PR related to linters:

• In harper-cli lint if after processing any --only and/or --ignore commandline arguments there are no lints configured, a warning will be issued. This is most commonly the result of making a typo when testing a new linter you're working on.

How Has This Been Tested?

Manually.

Checklist

• I have performed a self-review of my own code
• I have added tests to cover my changes

[elijah-potter]

I like the idea, but I don't love having these methods do IO of any kind. Would it be possible to refactor this into a test case on the curated LintGroup?

[hippietrail]

I like the idea, but I don't love having these methods do IO of any kind. Would it be possible to refactor this into a test case on the curated LintGroup?

Putting this straight in my inbox for today!

[hippietrail]

I like the idea, but I don't love having these methods do IO of any kind. Would it be possible to refactor this into a test case on the curated LintGroup?

I made it a test by adding an Option<Vec> to LintGroup for linter names that clash. This way I was able to keep references to LintGroup as they were.

I took the opportunity to refactor the curated linter constructor too by adding new macros to add linters that use either a dialect or dictionary. But not for SpellCheck which, for now, is the only one that takes both.

I did not remove any of the three clashing Linters. Up to you if you want to roll that into this PR when you test it, or do it as a separate PR after this one.

Until then this PR will show up as failing of course!