feat(commands): unescape yank-join separator

[RoloEdits]

Introduces a str module and an unescape function to helix-stdx, which processes escape sequences in strings and converts them into their corresponding literal characters. The function handles a variety of escape sequences, including:

• \n for newlines
• \t for tabs
• \u{...} for Unicode characters

The function does not unescape sequences like \\ to \, leaving them as they are. This opinionated behavior ensures that only certain escape sequences are processed, and is built around user input, not general input.

Given that its based around user input, a conservative approach was taken for handling bad input, where if the string cannot be processed as expected, it returns the original input.

Examples:

• Converting escaped newlines: unescape("hello\\nworld") results in "hello\nworld".
• Converting escaped tabs: unescape("hello\\tworld") results in "hello\tworld".
• Converting Unicode escape sequences: unescape("hello\\u{1f929}world") results in "hello🤩world".
• Handling invalid Unicode escape sequence: unescape("hello\\u{999999999}world") results in the original "hello\\u{999999999}world".

The implementation also includes tests, but no guarantees for edge cases.

Previously, the yank-join command joined the current selection with the separator as-is. With this update, escape sequences in the separator are unescaped to their corresponding literal characters before joining the selection, aligning with user expectations:

Newline:

Tab:

Unicode:

Closes: #10993

[RoloEdits]

In testing the user input I noticed that ' and " input by themselves, :yank-join ", would result in newlines. This exists on master as well, and seems to stem from how Shellwords::from handles these inputs, which results in the TypedCommand args being empty, and defaulting to the line_ending as a result. I thought about trying to tackle that here, but the scope would be beyond the current PR, so going to open this for review.