fix: raise the intended ValueError for non-ascii hex strings

[nikolauspschuetz]

hexstr_to_bytes guards padded_hex.encode("ascii") with except UnicodeDecodeError, but str.encode raises UnicodeEncodeError — so the handler never runs and a non-ascii hex string surfaces the raw codec error instead of the intended ValueError:

>>> HexBytes("0x0é")
UnicodeEncodeError: 'ascii' codec can't encode character '\xe9' in position 1: ordinal not in range(128)

Correcting the caught type raises the message the handler was written for — ValueError: hex string 0é may only contain [0-9a-fA-F] characters. Valid hex is unaffected. Added a regression test (there was none for non-ascii input).