fix unicode identifier parsing

[obfusk]

Correctly parse identifier that starts with a letter that is neither upper nor lower case (and should count as the latter); e.g. 猫.

I'm guessing this needs to be fixed in more places; I'd be happy to look into that sometime later, but I'm not very familiar with the codebase and would appreciate some help. I just found this specific bug when fixing a bug in hoogle.

I also did not add any extra tests; I can also do that later if you'd like.

[obfusk]

I suspect several other uses of isLower c should be isLetter c && not (isUpper c).

[DanBurton]

Instead of merging this, I've added a commit to handle it more thoroughly: c9f786d

And a test: 94a3bcd

There's still something weird about prettyprinting these kinds of identifiers, though. See also: #443

[obfusk]

Great! Thanks for saving me the extra work to handle it more thoroughly :)