Trie-based route parser

Suggested by #1889 

Currently, parsing is linear. We go through the route file one-by-one until we find a match. Additionally, we go through each repeated path segment multiple times- once for every time it is mentioned and reused. This is very slow.

Instead, we should create a trie datatype - `Trie (Piece Text)` - which can be used for more efficient parsing of the path pieces.