Test fails with Python 3.9

[swt2c]

There is a unit test failure with Python 3.9 (as observed on Fedora 33 and on Debian Unstable).

__ test_numeric_chars_contains_all_valid_unicode_numeric_and_digit_characters __

    def test_numeric_chars_contains_all_valid_unicode_numeric_and_digit_characters():
        set_numeric_hex = set(numeric_hex)
        set_numeric_chars = set(numeric_chars)
        set_digit_chars = set(digit_chars)
        set_decimal_chars = set(decimal_chars)
        for i in range(0X110000):
            try:
                a = chr(i)
            except ValueError:
                break
            if a in "0123456789":
                continue
            if unicodedata.numeric(a, None) is not None:
>               assert i in set_numeric_hex
E               assert 69573 in {178, 179, 185, 188, 189, 190, ...}

tests/test_unicode_numbers.py:49: AssertionError

Doing a little bit of debugging, indeed on Python < 3.9, unicodedata.numeric(chr(69573)) throws a ValueError, while on Python 3.9, it returns 1.0. Thus, it seems that new numeric hex value(s) have been added in Python 3.9? Is the solution to add 0x10FC5 to unicode_numeric_hex.py, but will that break things on older Pythons?

[SethMMorton]

Yeah, an update like this is expected with each Python minor release since new valid unicode characters get added - #104 was the merge request for when Python 3.8 was released.