Add ordinary_normalize

[saki7]

// Normalizes occurrences of "ordinary" inconsistent spelling with
// the canonicalized letter.
//
// This algorithm guarantees the character count never changes regardless
// of the transformation, in contrast to fully-featured Unicode normalization
// such as NFKC. (Note: NFKC not only changes the character count; it may
// occasionally render the resulting character to empty set, or even *flips*
// the character positioning, which makes it require a rather complex algo.)
//
// The above design choice enables this function to be used like an *easy*
// canonicalization, where the original and the transformed string can both
// be mapped to the identical `interval<int>`.

[cppwarningnotifier]

Environment				C++23	C++26
iris	Clang	22	Debug	✅success	✅success
			Release	✅success	✅success
	GCC	15	Debug	✅success	✅success
			Release	✅success	✅success
	MSVC	2026	Debug	✅success	✅success
			Release	✅success	✅success