Move to purescript-parsing?

[newlandsvalley]

Because it's supposed to be quicker. Also, we should be able to pre-compile all the regexes (of which there are many) and so obviate the need to re-compile at each invocation.

[newlandsvalley]

This has now been implemented in the parsing branch.

[newlandsvalley]

Unfortunately, benchmarking shows this to be nearly twice as slow as the original string-parsers version. Perhaps the effort of placing all pre-compiled regexes into a Map is too expensive?

[newlandsvalley]

Using HashMap rather than Map for the regex map makes hardly any difference. Removal of HashMap altogether (and re-introducing regex re-parsing) improves speed a good deal, but is still about 16% slower than the string-parsers version. This version of the code is pretty much a direct replacement of string-parser with parsing with little or no structural changes.

This suggests that it is not appropriate to replace string-parsers but perhaps we can make small speed improvements by replacing gratuitous regexes which have multiple invocations in a parse - particularly those invoked during a parse of the tune body.