Convert clipboard line ending to document line ending when pasting

[kirawi]

Resolves #715

WIP because it doesn't play nice with the LSP.

[cessen]

Yeah, I think what we actually want to do here is have a stand-alone function that (probably?) takes a &str and a target line ending, and returns a String with all non-specialized line endings replaced with the target line ending. Then we can just call that from the paste command to do the conversion before actually pasting.

The non-specialized line endings are CRLF, LF, CR, LS, and Nel, and should be replaced as needed.

(FF, VT, and PS have special uses, and shouldn't be replaced. Incidentally, I just noticed I screwed that up here:

helix/helix-term/src/commands.rs

Lines 1427 to 1435 in fe17b99

	cx.editor.set_status(match line_ending {
	Crlf => "crlf".into(),
	LF => "line feed".into(),
	FF => "form feed".into(),
	CR => "carriage return".into(),
	Nel => "next line".into(),
	// These should never be a document's default line ending.
	VT | LS | PS => "error".into(),

Ha ha. So we should probably put what counts as "specialized" or not somewhere centralized, so we only have to get it right once. Probably just a const &[] of the relevant line endings will do, so we can call contains())

[cessen]

This prompted me to investigate the purpose of the various line endings further, which led me to this:
https://www.unicode.org/standard/reports/tr13/tr13-5.html

It's actually not entirely clear to me yet which line endings should be treated as "specialized" or not. So let's definitely centralize that, and then we can adjust as needed later. (I'm suspecting that everything except CRLF, LF, CR, and Nel should probably be treated as specialized, but I'm not sure.)

[kirawi]

Perhaps we could use memchr to check for the line endings? It's already a dependency. Or perhaps regex.

[archseer]

for the clipboard one:

    // only compile the regex once
    #[allow(clippy::trivial_regex)]
    static REGEX: Lazy<Regex> =
        Lazy::new(|| Regex::new(r"\r\n|\r|\n").unwrap());
    REGEX.replace_all(paste_string, new_line_ending)

[archseer]

I'd completely drop the line ending conversion from this PR and only land the clipboard change. Fixing document-wide line endings is simple to do from the terminal, we can implement a better solution later.

[cessen]

Lazy::new(|| Regex::new(r"\r\n|\r|\n").unwrap());

Yeah, for now I think that's good. We know at least those should be replaced, and if/when I get around to more fully investigating the Unicode line ending situation I'll make a PR to include any others that should be covered as well.