can/should we switch from `msvcrt.getch()` to `msvcrt.getwch()`

[Cube707]

getwch() directly returns a string, so we could remove the manual decoding done here:

python-readchar/readchar/_win_read.py

Lines 10 to 11 in 9991a0b

	# manual byte decoding because some bytes in windows are not utf-8 encodable.
	return chr(int.from_bytes(msvcrt.getch(), "big"))

Does a swicht pose problems? Can we overcome theme?
Are there donwsides?

[Cube707]

Problem 1: using the corisponding ungetwch() in the tests raises a bounch of OSError's