The regex code has historically returned the actual matched patterns and the
named capture groups within regexes. When trying to count common and/or large
patterns, this leads to very large outputs.
I've added two new functions -RPc and -CPc that will cause wikiq to return
counts of each pattern (0 when there are no matches). The options apply to all
comment or revision patterns. I considered interfaces to make it possible to do
some but others but concluded this would be too complicated an interface.
This code should be checked before it's merged.