feat: new ReedlineEvent::PartialComplete event

[blindFS]

Move the partial completion logic from the handling of ReedlineEvent::MenuNext, so that we can fix nushell/nushell#14152 while keeping the following behavior.

When the only valid options are say foo and foooo, a typical bash behavior of f#tab would complete the repl text to foo without moving the menu item towards foooo.

Closes #881

[blindFS]

I'm not very sure that it's safe to remove this. But I think it's a bad idea to handle quick completion related menu updates here.

[fdncred]

Are there tests we can add for this plus the quick complete section that was removed?

[blindFS]

Are there tests we can add for this plus the quick complete section that was removed?

That's a good question.

1. I can't find any unit test for reedline events, seems that it is the function behind each event handler that's being tested.
2. As for the removed part, I've changed this macro (same for ide_menu.rs) so that existing tests for partial completions shall pass. But I don't think that's good enough. The fact that we have to call menu.update_values twice (one before the can_partial_complete check, one inside it) for partial_completion tasks bugs me. I don't know if I can help to optimize it, but that's probably off-topic.

reedline/src/menu/columnar_menu.rs

Lines 773 to 795 in 22768f3

	macro_rules! partial_completion_tests {
	(name: $test_group_name:ident, completions: $completions:expr, test_cases: $($name:ident: $value:expr,)*) => {
	mod $test_group_name {
	use crate::{menu::Menu, ColumnarMenu, core_editor::Editor, enums::UndoBehavior};
	use super::FakeCompleter;
	$(
	#[test]
	fn $name() {
	let (input, expected) = $value;
	let mut menu = ColumnarMenu::default();
	let mut editor = Editor::default();
	editor.set_buffer(input.to_string(), UndoBehavior::CreateUndoPoint);
	let mut completer = FakeCompleter::new(&$completions);
	menu.can_partially_complete(false, &mut editor, &mut completer);
	assert_eq!(editor.get_buffer(), expected);
	}
	)*
	}
	}
	}