This PR adds functionality to skip around ANSI escape sequences in catch_textflow so they do not contribute to line length and line wrapping code does not split escape sequences in the middle. I've implemented this by creating a AnsiSkippingString abstraction that has a bidirectional iterator that can skip around escape sequences while iterating. Additionally I refactored Column::const_iterator to be iterator-based rather than index-based so this abstraction is a simple drop-in for std::string.

Currently only color sequences are handled, other escape sequences are left unaffected.

Motivation: Text with ANSI color sequences gets messed up when being output by Catch2 #2833.

Closes #2854

`max_digits10` is the number of decimal digits that the type _can_
represent, in other words, the number of decimal digits needed to
disambiguate any two floating point numbers when serialized to
a string.

Closes #1210

When a signal is caught, the destructors of Sections will not be called.
Thus, we must call `sectionEndedEarly` manually for those Sections.

Example test case:
```
TEST_CASE("broken") {
   SECTION("section") {
      /// Use illegal cpu instruction
      __asm__ __volatile__("ud2" : : : "memory");
   }
}
```

Remove `#pragma intrinsic(_umul128)` because it doesn't work on
ARM64EC and x64 works without it.

Co-authored-by: Agoston Szepessy <agos@microsoft.com>

The WithinRelMatcher does not listen to the precision specification; it just naively pipes to stringstream. If you specify the precision, that has no impact on the outputted values when the match fails.

This fix makes the WithinRelMatcher listen to the precision (https://github.com/catchorg/Catch2/blob/devel/docs/tostring.md#floating-point-precision) specified by: Catch::StringMaker<float>::precision

Apparently I looked at the docs for umulh when checking availability,
and umul128 is x64 only.