Validate that the new "Toggle Block Comment" command correctly removes itself

[alex-bochkov]

I noticed some unusual behavior inside the Citrix VDI a few weeks ago where characters were being removed from incorrect positions, but I haven't been able to reproduce it since. It's possible the issue was related to the VDI itself and was resolved after a restart.

@Blake-goofy , could you please give this feature a quick test when you have a few minutes? Nothing extensive is needed, just a general sanity check.

[Blake-goofy]

I can take a look this weekend.

[alex-bochkov]

I'd appreciate it if you could handle the release as well. I'm completely swamped through mid-July.

[Blake-goofy]

I actually found some time this morning. Everything I threw at it seemed to work as expected.

I can handle the release.

[alex-bochkov]

Something is definitely broken here. It fails, but not consistently. I need to review the code and see what Codex has actually done.

[Blake-goofy]

I found a real scenario where it messes up

-- not a block comment: /*
SELECT 1; /* real comment */

Put the caret after the /* on the first line comment, then run the command; result:

-- not a block comment: 
SELECT 1; /* real comment *

[Blake-goofy]

Fixed in 02d8f03.

• If text is selected, behavior is unchanged: wrap/unwrap selected text.
• If nothing is selected, it now tokenizes the SQL and only acts when the caret is inside a real TSqlTokenType.MultilineComment.
• It ignores /* inside -- line comments, string literals, bracketed identifiers, etc.
• It handles nested block comments by removing the innermost enclosing /* */.