Fix bug in parsing hex char escapes

[RadixSeven]

Bugfix: the parser would crash if it encountered "\x5C" in a grammar. This commit fixes that bug. (There is still a bug I couldn't locate that causes the parser to crash with "\\". But the "\x5C" provides a workaround.)

The bug was that the parsing code returned an int instead of a str, as the client code expected.

This commit also adds some test cases I used to find the problem.

[Saibo-creator]

@RadixSeven Thanks for your contribution! The fix seems good to me :)

Just to clarify before we proceed with the merge:

There's still an elusive bug that causes the parser to crash when using "\". However, using "\x5C" seems to be a viable workaround.

Could you provide a specific example where the parser fails? I'm not clear on that.

The parser functions correctly with both:

root ::= ("\x61" "\x5C")*

and

root ::= ("\x61" "\\")*

The output, a\a\a\a\a\a\a\a\a\a\a\a\a\a\a\a\a\a\a\a, appears as expected.