Fix wildcard hx-on search's root node

[Telroshan]

Description

As @svenberkvens pointed out in #2019, the XPATH query run to find hx-on wildcard elements to process, as well as its IE11 compatible code counterpart, didn't take the root node parameter into account, resulting in scanning the whole DOM to find such elements.
This was causing a huge performance impact for each individual node processed later on by htmx on a initially large DOM.

This PR fixes this issue by looking for hx-on wildcard elements from the root element parameter, instead of from the document itself.

Corresponding issues: #2019 and #2051

Testing

Ran a performance test using htmx version before the introduction of that hx-on wildcard attribute (i.e. 1.9.2), one with the current version, and another one with this PR's changes
The test creates 100 000 links in the DOM, then adds a dummy element and calls htmx.process on it

• In this JSFiddle using 1.9.2, I get the following:

Took 7.3 ms to process node

• In this JSFiddle using 1.9.9, I get the following:

Took 591 ms to process node

• In this this JSFiddle using this PR's changes, I get the following:

Took 7.9 ms to process node

So it looks like we're back to the initial performance of processing an individual node with a very large DOM around.

Also ran the test suite locally which works, no no regression seems to be introduced by this change

Checklist

• I have read the contribution guidelines
• I have targeted this PR against the correct branch (master for website changes, dev for
  source changes)
• This is either a bugfix, a documentation update, or a new feature that has been explicitly
  approved via an issue
• I ran the test suite locally (npm run test) and verified that it succeeded

[alexpetros]

I'm not convinced that that any of the shouldProcessHxOn part is necessary. Doesn't the performance impact primarily come from the lines 1921 and 1924?

[alexpetros]

Could you run your test suite against the two-line version of this PR, that only makes the changes on lines 1921 and 1924?

[Telroshan]

@alexpetros Running the xpath query from the element itself with that leading . syntax unfortunately doesn't process the root node itself. Same goes for getElementsByTagName in the IE fallback code that will get all children, but not the root node itself, thus this first check to add the root itself to the array if it should be processed for hx-on attributes.

Removing this if breaks a lot of hx-on tests btw

[Telroshan]

Then the function is just there to avoid duplicating the logic from the IE fallback code + the root node check

[alexpetros]

Oh I understand you moved it—I didn't see the delete bit below.