nl: implement -d/--section-delimiter

[cakebaker]

This PR implements -d/--section-delimiter that allows you to define a section delimiter other than the default \:. If the specified delimiter is a single character, : is added to the delimiter.

[tertsdiepraam]

Alright, so this is more of a brainstorm than actual feedback, because this expression is very cool and clever but also a bit weird. Maybe it could be:

if s.trim_left(self.delimiter).is_empty() {
    match s.len() / self.delimiter.len() {
        3 => SectionDelimiter::Header,
        2 => SectionDelimiter::Body,
        1 => SectionDelimiter::Footer,
        _ => SectionDelimiter::None,
}

This avoids the allocations of the strings from repeat as well. But I'm not actually sure it's better.

[tertsdiepraam]

Does making self.delimiter an Option<SectionDelimiter> make sense? Or maybe the whole SectionDelimiterParser?

[tertsdiepraam]

Suggested change

	let delimiter = if delimiter.chars().count() == 1 {
	let delimiter = if delimiter.len() == 1 {

[cakebaker]

Hm, len returns the number of bytes, not the number of chars (for example, "ä".len() returns 2). And my intention was to get the number of chars. On the other hand, using len leads to the same output as GNU nl when specifying ä as delimiter. Which is incorrect, in my opinion ;-) I guess I have to ask in the GNU coreutils project what the expected output should be when using a single non-ASCII char.

[tertsdiepraam]

Ohh I see. Maybe it should even be unicode width?

[cakebaker]

@tertsdiepraam I addressed your feedback in the latest push in the following way:

• I replaced chars().count() with len() and added an additional test case test_non_ascii_one_char_section_delimiter
• I rewrote the "weird" match (plus moved the entire function to SectionDelimiter)
• the parse function now returns an Option<SectionDelimiter>

In addition I got rid of the SectionDelimiterParser.

[tertsdiepraam]

This is beautiful now!

[tertsdiepraam]

CI failures seem unrelated, but I'm rerunning them just to be sure.

[cakebaker]

Somehow I overlooked that this PR got approved a while ago :|