Moving from pre-commit to hk/lefthook + mise

[scop]

I've moved from pre-commit to the Lefthook+mise combo for a lot of my projects, and found it working better in almost all respects, and would suggest us to move toward that direction in bash-completion too.

Performance/parallelism config, having correct version of tools installed also for general use as opposed to just when ran through pre-commit, no python required (python is not a problem for bash-completion though due to our test suite), customizability/overridability for different environments, ability to use a lot of tools without them needing a upstream pre-commit config somewhere are some of the benefits of the move that spring to mind right now.

A slight drawback is some more/verbose config required, and some limitations in available tool install options (even though in general there are more available options with this combo) that might affect bash-completion's perl checker things.

Along with this change I'd like to move from gitlint to committed, pymarkdown to dprint (which brings in a bunch of new formatters too). This is something that could likely be easily done while we're with pre-commit still, if wanted.

Namedropped also hk in the title. I've no experience with it yet, but it does sound interesting as an alternative to Lefthook.

No rush with this, just to drop the thoughts out there.

[akinomyoga]

I don't have a particular preference on this. I've used pre-commit just because it's been provided by bash-completion. If it switches to new frameworks, I can learn them.