Implement measure_text_width with no allocation

[remi-dupre]

I implemented that while working on optimizing something I don't remember, it might still be useful.

Also, I noticed that the test for this function was wrong when only using the unicode-width feature (probably due to some update in the corresponding?).

# Running on origin/main
$ cargo test --no-default-features -F unicode-width 
running 6 tests
test utils::test_pad_str ... ok
test utils::test_truncate_str_no_ansi ... ok
test utils::test_pad_str_with ... ok
test utils::test_text_width ... FAILED
test utils::test_attributes_many ... ok
test utils::test_attributes_single ... ok

failures:

---- utils::test_text_width stdout ----
thread 'utils::test_text_width' panicked at 'assertion failed: `(left == right)`
  left: `21`,
 right: `17`', src/utils.rs:940:5
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

I've completed the Makefile so that all possible sets of feature are tested, as the project is not too big I don't think it's too heavy.

[djc]

Thanks, this mostly looks good!

[djc]

Nit: suggest spelling it like this:

.filter_map(|(s, is_ansi)| (!is_ansi).then(|| str_width(s)))

[remi-dupre]

I find it less readable, but that's your project 👍

[remi-dupre]

Actuallyyyy, clippy doesn't like it either! 🤓

https://rust-lang.github.io/rust-clippy/master/index.html#filter_map_bool_then

[djc]

Fair enough. In that case, I'd like it to be written like this (assuming clippy is okay with it):

.filter_map(|(s, is_ansi)| match is_ansi {
    false => Some(str_width(s)),
    true => None,
})

[remi-dupre]

Clippy is okay with that 👍

[remi-dupre]

I'll investigate on why the tests fails later, I start to realy suspects some behavior change in unicode_width (I know this sounds cliché but it worked on my computer 😢)

[remi-dupre]

Tests are fixed.

This was just a rooky mistake of forgetting to put .force_styling(true) in the new test case, it took me way too much time to figure that out 😅

[djc]

Nice, thanks!