This PR improves the code in round_digits: now the first digit in the stddigits list is excluded from the list of digits that must be rounded up when the round_up rounding mode is used. When rounding to a 0 digit, all the digits following it are looked up, and if one of them is different from 0, the 0 is rounded up.

This fixes some wrongfully formatted floats as, for example, when running the following code in the master branch

>>> from gmpy2 import mpfr
>>> from mpmath import mp
>>> print('{:.2Uf}'.format(mpfr('0.25')))
0.25
>>> print('{:.2Uf}'.format(mp.mpf('0.25')))
0.26
>>>