.downcase(Unicode::CaseOptions::Fold) using char ranges

[akzhan]

The Fold option is required for correct Unicode string comparison using fold case.

Reimplementation of #4511 using char ranges.

Less memory used.

[asterite]

Use a tuple to avoid a memory allocation:

return {char.ord + result} if result

[akzhan]

String.compare will be reworked by another PR

[akzhan]

Is it GTG?

I need these changes to continue Unicode support.

Travis build is broken somewhere unrelated to PR.

[asterite]

@akzhan I'm not sure why this is useful. Where would one use it? Plus, if it's to solve another issue (the one about unicode normalization) I would send a single PR targeting that. Is there any other language where you can downcase and specify to fold case?

[akzhan]

@asterite full Unicode case mapping added in Ruby 2.4 with String.downcase(:fold).

Take a note that case folding is not usual normalization, see Section 3.13 of the Unicode standard.

Also X.casefold() == Y.casefold() in Python 3 implements the "default caseless matching" (D144).

Real case fold comparison doing like:

def NFD(text):
    return unicodedata.normalize('NFD', text)

def canonical_caseless(text):
    return NFD(NFD(text).casefold())

[akzhan]

String.normalize will be subject of another PR.

[asterite]

@akzhan Thank you! I'll review this today or during this week and probably merge it.