Make sure that jsdoc is always in range with parameter declaration

[sheetalkamat]

Fixes #56677

[weswigham]

Funky. I'm not sure why we even parse a comment range on the incremental reparse in this case, since we deleted the leading /. The comment range scanner function pretty clearly only matches comments on seeing a // or /* sequence. The single / left after the edit should be interpreted as a division operator or opening to a regex literal, I'd think. I get that the jsdoc nodes we make from these spans, and adjusting them in the incremental parser, is ultimately what triggers the assertion (which is why excluding them fixes it), just feels weird that we have a jsdoc node at all after the edit - I'd assume we'd have rescanned from in front of the jsdoc host node and thrown it out.

[jakebailey]

I guess I would have thought this were a bug somewhere in updateTokenPositionsAndMarkElements in the incremental parser, since we should be marking nodes as need to be fully reparsed...

[sheetalkamat]

The issue was that when first source File was created with function bug(\r\n test /** */ true = test test 123\r\n) {} it created a parameter declaration of 0 width just before the jsdoc and jsdoc was attached to that parameter and hence the assert in incremental parsing was failing where it adjusts the spans for nodes in the change range and keeps it around so if parser can reuse node, it will. ( in this case it obviously doesnt as removing that leading / makes it expression)