[INTT-54] Add code formatters for general projects

[tlater-famedly]

This adds formatting with treefmt, including a pre-commit hook for it. This seems to be reasonably performant, so I think it's fine to keep it in the pre-commit hooks.

Documentation is added via devshell commands, which should hopefully be convenient enough for downstream users.

[RasmusRendal]

We have programs.shfmt.indent_size = 0; to use tabs for indentation :))

[tlater-famedly]

That's the default setting:

  -i,  --indent uint       0 for tabs (default), >0 for number of spaces

[RasmusRendal]

We explicitly went without sqlfluff, because it couldn't handle our complicated postgres

[RasmusRendal]

Might want to borrow our prettierrc:

{
  "overrides": [
    {
      "files": [
        "*.yml",
        "*.yaml"
      ],
      "options": {
        "singleQuote": false,
        "printWidth": 120,
        "tabWidth": 2,
        "proseWrap": "preserve"
      }
    }
  ]
}

[RasmusRendal]

arguing about formatting timeeeeee.

Maybe one formatter per PR would be the way to go :(

[RasmusRendal]

This is pretty great

[emgrav]

Looks good! I have two questions that are probably not really relevant to this PR, but if the blockers are not big it would be cool if we were able to close these tickets with this PR if we can clear the blockers.

[emgrav]

What exactly is needed to make rustfmt run as a pre-commit hook after trailing-whitespace? Adding this back in is tracked in https://famedly.atlassian.net/browse/INTT-55
If we already know what the blocker preventing us from re-enabling it is, it would be great if this information could be added to the ticket :)

[RasmusRendal]

The only thing blocking trailing-whitespace running after the formatter on main is that on main, there's no formatter.

[tlater-famedly]

Yeah, the reason it wasn't as of this PR is because we picked it up in nehws CI. It's a rustfmt bug to begin with, nehws just couldn't start using the pre-commit hooks if we left this one in.

We should probably just fix it as part of this PR.

[emgrav]

Same thing here, we are tracking the missing check-yaml in INTT-56. In what way does it need to be tweaked for helm charts?

[RasmusRendal]

It probably needs to be disabled entirely. Last time I checked, check-yaml breaks down completely in the face of go templating.

[emgrav]

Could we then maybe standardize around keeping helm charts always in a top-level directory called charts? Then we could reenable check-yaml and simply except crates/** from the check

[tlater-famedly]

I think I like that better than permanently omitting this check. It's nice to be certain that any YAML files that make it into our repos evaluate correctly.

If Helm charts are simply not valid YAML we should find a way to exclude them.