Gate Format Selections on whether the active formatter can actually format ranges

[prertik]

What Changed

• compute range-format support from formatter configuration and language-server capabilities
• show Format Selections only when at least one selected buffer has a range-capable formatter
• keep Prettier-supported range formatting available without depending on LSP support
• hide the action for formatter modes that cannot honor selections, such as external-command and code-action formatters
• keep the action handler safe by rechecking support and returning early when range formatting is unavailable
• restrict capability checks to the servers actually registered for the current buffer
• add regression coverage for supported, unsupported, and mixed-server cases
• document when Format Selections is available in both action docs and user docs

Self-Review Checklist:

• I've reviewed my own diff for quality, security, and reliability
• Unsafe blocks (if any) have justifying comments
• The content is consistent with the UI/UX checklist
• Tests cover the new/changed behavior
• Performance impact has been considered and is acceptable

Closes #25796

Release Notes:

• Improved Format Selections so it is only shown when the active formatter supports range formatting.

Testing I did:

Here are the formatters I used:

Note: I've chosen three formatters:

1. Prettier, which supports range formatting
2. clangd (LSP) which also supports range formatting
3. gopls which do not support range formatting

Screencast_20260405_102431.webm

[prertik]

@SomeoneToIgnore Please review.

[SomeoneToIgnore]

Sorry, test_sidebar_invariants is bad and is going to be fixed by teammates, hopefully soon, this will require another rebase later.

[prertik]

Sorry, test_sidebar_invariants is bad and is going to be fixed by teammates, hopefully soon, this will require another rebase later.

Thanks, @SomeoneToIgnore. I was wondering why tests are failing all of a sudden and trying to look into issue without much luck.

[SomeoneToIgnore]

Seems to be fixed?
#53421

[SomeoneToIgnore]

Thank you, the way to disable the action seems way better than my proposal, but the implementation is quite odd.

Let's clean it up, as it looks we can do it much simpler.

[prertik]

Let's clean it up, as it looks we can do it much simpler.

Hi @SomeoneToIgnore, I've cleaned it up based on your suggestions.

[SomeoneToIgnore]

I've pushed a few fixes as there was a lot of unneeded allocations all over the place.
There's one very odd place left that we need to discuss.

[SomeoneToIgnore]

Sorry, lsp_store is a very important module and I want to understand what those changes are meant for: I would appreciate an elaboration on what's going on and the purpose of it all, given the #53178 (comment) agreement that the old code was working ok for our needs.

[prertik]

Sorry, lsp_store is a very important module and I want to understand what those changes are meant for: I would appreciate an elaboration on what's going on and the purpose of all it, given the #53178 (comment) agreement that the old code was working ok for our needs.

Apologies, I agreed to the change, but my commits during rebase, I added wrong changes and committed them. I'll amend and push again.

[SomeoneToIgnore]

Thank you.

Given all the back-and-forth, I'll merge this after the next release this Wednesday, to be on a safe side.

[prertik]

Thank you.

Given all the back-and-forth, I'll merge this after the next release this Wednesday, to be on a safe side.

Sure, @SomeoneToIgnore. Thanks for reviewing this and giving me feedback.