feat(lint): add useGoogleFontPreconnect rule

[kaioduarte]

Summary

Implement google-font-preconnect from eslint-plugin-next.

Test Plan

Snapshots + CI

[codspeed-hq]

CodSpeed Performance Report

Merging #4185 will not alter performance

_{Comparing kaioduarte:google-font-preconnect (eeb1d24) with main (6c209ef)}

Summary

✅ 101 untouched benchmarks

[ematipico]

@kaioduarte a suggestion for this and other PRs. When filling out the template, please tell how you named the rule inside Biome.

Since Biome has a strict naming convention, it's often difficult to understand which rule you implemented.

[ematipico]

The diagnostic must match our rule pillars

[ematipico]

Thank you! I suggested few changes. Once addressed, we can merge the PR

[ematipico]

Is this check needed? The check was already done in run. Did I miss something?

[kaioduarte]

Yes, I didn't want to change the value if a rel already exists, maybe we could do so, I'm not sure :)

The action is a bonus since the original rule doesn't have one.

[ematipico]

Suggested change

	let rel = node.get_attribute_inner_string_text("rel");
	if href.starts_with("https://fonts.gstatic.com") && (rel.is_none() || rel? != "preconnect")
	let rel = node.get_attribute_inner_string_text("rel")?;
	if href.starts_with("https://fonts.gstatic.com") && rel != "preconnect"

Since rel is mandatory attribute, we can simplify the check

[kaioduarte]

It's not mandatory, we emit a diagnostic in the absence of the property or if its value is different of preconnect.

[ematipico]

Then you shouldn't use rel?, because the try operator ? exit from the function in case is None, which goes in conflict with the accepted value rel.is_none. I suggest using matches!(rel, Some("preconnect")) instead.

While it's true we're using an OR operation, it would fail if you flip the arguments