I noticed the following TODO in io.undertow.util.PathTemplateMatcher:

TODO: we can probably do this faster using a trie type structure, but I think the current impl should perform ok most of the time

The objective of this ticket is to develop a faster path template matcher / handler.  

Initially I attempted to use a tree type structure to develop a faster router, but the code evolved into something a little more complex.  Basically the matcher - called router in order to temporarily run it side-by-side with the original matcher - consists of strategies that optimised to quickly match specific combinations of path segments.  

Based on - admittedly rudimentary - performance benchmarks, the new version appears to be between 3 and 4 times faster than the previous version, depending on the number of paths that needs to be recognised.  

There are one or two more minor improvements I would like to make before submitting a PR.

PS: This is my first attempt at contributing towards this project, so there may be several things about contributing to this project that I don't know.  Feel free to point them out to me!