Get line and character numbers from parser and add to inference

[emptyflash]

Pretty sure megaparsec has info about what line and character number a given expression came from. We could add this to the expression ADT so we could provide more helpful unification and other inference errors.

mrkkrp/megaparsec#343
https://stackoverflow.com/questions/54430335/errorbundles-after-parsing-with-megaparsec

[emptyflash]

This is "done". I think there are some code paths not handled, but I think I covered most of them in 3c83d69 and f9c3b1b