Introduce prepend_to_selector() to avoid additional if checks and follow single responsibility principle

[felixarntz]

What?

This implements the WP core PR WordPress/wordpress-develop#4380 in Gutenberg first, as requested on that PR, which makes sense given the current Gutenberg implementation differs slightly from current WP core's.

Why?

Having to check on every append_to_selector() call what the append position is inefficient and violates the single responsibility principle. The code that calls this method already knows whether it needs to append or prepend, so having a separate prepend_to_selector() method with only that responsibility is more efficient.

Related: #47833

[github-actions]

Flaky tests detected in 45473d9.
Some tests passed with failed attempts. The failures may not be related to this commit but are still reported for visibility. See the documentation for more information.

🔍 Workflow run URL: https://github.com/WordPress/gutenberg/actions/runs/4866147232
📝 Reported issues:

• [Flaky Test] should apply Heading block styles to all Heading blocks #49942 in /test/e2e/specs/site-editor/push-to-global-styles.spec.js

[costdev]

As noted in a Slack DM to @felixarntz:

A $to_prepend of ".some-class " will produce: .some-class h1,.some-class h2,.some-class h3.

This was the result prior to this PR though. See this 3v4l.

One way to achieve the result in the docblock is trim( $sel ) and implode( ', ', $new_selectors ). See this 3v4l.

I'm not sure if these "misplaced" spaces were already known and accepted, or if this is something tests should have caught.

[felixarntz]

Thanks @costdev, good catch!

We could add trim() to the function, but to me that seems like a separate conversation. The output with trim would be preferable, but it's also not a big deal to have the extra space, and since the function is called tons of times there may be some (probably insignificant) performance cost to it.

In other words, I'm by no means opposed, but potentially this could be discussed in a separate issue / PR.

I have fixed the doc block to include the double spaces to clarify that.

[costdev]

Sounds like a plan @felixarntz!

For a separate issue / PR, as an alternative to the trim(), if we know that the format always includes a space after the comma, we could explode( ', ', $selector ) and implode( ', ', $selector ). See this 3v4l.

[spacedmonkey]

@felixarntz Coding standards failure needed to be fixed.

[spacedmonkey]

Note: we should backport this - #47833 as the same time.