Unbased unsized literals in BH

[matx-chris]

This PR adds the '0 and '1 syntax for unbased unsized literals to the Bluespec Haskell parser.

Notes:

• ' is an identifier character, so a literal like this must start after a token boundary; i.e., foo'0 is an identifier, but foo '0 is a function applied to an unbased unsized literal.
• '0' is still a character literal
• This is similar to a lot of Haskell syntax for TH and Data Kinds promotion, it doesn't actually conflict with any valid Haskell syntax, which is nice.

[nanavati]

LGTM, my only question is how does this compare to what BSV does with '0 ad '1?

[matx-chris]

LGTM, my only question is how does this compare to what BSV does with '0 and '1?

I believe it does the same thing. I followed the same implementation. They both produce cVar (idConstAllBitsUnsetAt pos) and cVar (idConstAllBitsSetAt pos) in the parser. There's a slight difference in 'x and 'z, which are accepted by the BSV lexer, but then rejected in the parser. Here, I didn't accept them in the lexer either. The reason is that I wanted to minimize the syntax grab.

[matx-chris]

I realized I missed one case, which is '0 as a pattern. (Note: '1 as a pattern is an error in BSV as well, so this only applies to '0.) Added that as well, for consistency.

[quark17]

I realized I missed one case, which is '0 as a pattern. (Note: '1 as a pattern is an error in BSV as well, so this only applies to '0.) Added that as well, for consistency.

I don't know why '1 in pattern matching is an error in BSV -- note that this is Issue #243.

[matx-amy]

reviewed, 👍

[rsnikhil]

Looks good to me.

[rsnikhil]

Approved (comments, if any, in PR comment list)