* replaced the implementation of the search index. the search index now uses a search-trie. edges have gates that can be used to check if a character can be matched when following the edge.
* SearchMatches now have a text and a list of NameHierarchies that share this node (instead of Ids, so that nodes must be found by names (which makes sure that SearchMatches are still valid after refresh).
* when parsing a project's description the CodeController now looks for NameHierarchies instead of SearchNode names to find ids of linked elements.