ptx: use char count instead of byte index to handle utf-8 characters

[aaron-ang]

close #2049

As discussed in the issue, ptx breaks when handling utf-8 characters (which may span multiple bytes). We need to use character positions instead of byte positions to index into chars_line. We can do so using .chars().count(). I created a helper function to consolidate this change since format_roff_line and format_tex_line use the same logic. I've also added some comments and renamed variables for readability.

[aaron-ang]

printf "it’s disabled\n" > foo.txt

## in main
cargo run -q ptx -G foo.txt
# thread 'main' panicked at src/uu/ptx/src/ptx.rs:610:32:
# slice index starts at 15 but ends at 13
# note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
cargo run -q ptx -G -T foo.txt
# thread 'main' panicked at src/uu/ptx/src/ptx.rs:566:32:
# slice index starts at 15 but ends at 13
# note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

## in ptx-panic
cargo run -q ptx -G foo.txt
# .xx "" "it’s" "disabled" ""
# .xx "" "" "it’s disabled" ""
cargo run -q ptx -G -T foo.txt
# \xx {}{it’s}{disabled}{}{}
# \xx {}{}{it’s}{ disabled}{}

## system
ptx -G foo.txt
# .xx "" "it’s" "disabled" ""
# .xx "" "" "it’s disabled" ""
ptx -G -T foo.txt
# \xx {}{it’s}{disabled}{}{}
# \xx {}{}{it’s}{ disabled}{}

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/timeout/timeout (fails in this run but passes in the 'main' branch)

[cakebaker]

Thanks!