feat(linter): Support potentially dangerous fixes behind a --fix-dangerously flag

[DonIsaac]

Some problems are potentially auto fixable, but doing so may be undesireable. In some cases, it could produce incorrect code. In other cases, it could change the behavior of applications.

ESLint handles this with Rule Suggestions, which are made available to language servers but do not get applied via --fix.

Since the linter is mostly used as a CLI (oxlint), we need to support applying suggestions there too. I think we should do this with a --fix-dangerously flag.

Tasks

NOTE: only the first two tasks are prerequisites for #4179

• refactor(linter): update FIx to support suggestions
• feat(cli): add --fix-dangerously flag
• feat(lsp): add command to apply suggestions

[Boshen]

Interesting, can you hold onto adding the flag and the command? I want to research and understand these first.

[DonIsaac]

Too late unfortunately, I've got a draft out already. I'll avoid merging it until you confirm though.

[DonIsaac]

We may also want to support multiple suggestions for each diagnostic.