Reject overlong UTF-8 forms in _Mbrtowc and add test case

[cpplearner]

This PR adds test case on top of #5943, and removes the check for out-of-range code points.

While _Mbrtowc is used by both <locale> and <format> facilities, only the former are affected because the <format> functions call the constexpr function _Decode_utf if the encoding is UTF-8.

I removed the check for out-of-range values because I think wch can't exceed 0xFFFF: there's an early return if the result won't fit in the Basic Multilingual Plane,

STL/stl/src/xmbtowc.cpp

Lines 110 to 115 in 3f76681

	if (trailingUtf8Units > 2) { // this would result in a UTF-16 surrogate pair, which we can't emit in our
	// singular output wchar_t, so fail
	// see N4950 [locale.codecvt.virtuals]/3
	errno = EILSEQ;
	return -1;
	}

[StephanTLavavej]

Thank you so much for figuring out how to salvage this! 😻 Now I finally understand what the issue is. I pushed additional test coverage.