Safer query selector

[toasted-nutbread]

This change adds a querySelectorNotNull function which does a null check. This is used in places where a JSDoc annotation type cast would be performed, which is generally undesirable and somewhat unsafe. While this still makes some assumptions about the actual type of the Element, and it often casts it to something else such as HTMLElement unsafely, it will throw if a null is encountered.

A few situations where null is permissible have been addressed, a few instances of ancient code have been removed, and a bunch of querySelector field initializations in the settings scripts have been simplified.

[github-actions]

✔️ No visual differences introduced by this PR.

View Playwright Report (note: open the "playwright-report" artifact)

[djahandarie]

Thank you, this looks much nicer. I notice this PR moves DOM operations from prepare to the class constructor in a number of cases. It certainly seems less redundant, but I wonder what was the intent of splitting it like that in the first place? Are there cases where the DOM might not be ready yet in the constructor?

[toasted-nutbread]

In general, there shouldn't be any situations where they weren't already existing, since prepare is called immediately after constructor in most cases. I think it was just as dumb design decision by me when I was doing that. I try to keep constructors free of side effects, but querySelector isn't really a side effect so I'm not sure why I did that for a bunch of things.