ED now faster and lighter with json

Switched from sending XML renditions of term collections to json objects. Loads faster and helps quite a bit with the memory problems.

Still a bit slow with the bigger terminologies (e.g. GO biological process ontology), but think we are getting pretty close to optimal given the current architecture. Suspect a move to a new independent website will probably be needed to make this work with the bigger vocabs (wordnet, UMLS, NCI,..) - that will likely be accessed via some ajax/web service combo.

- though it may also be possible and beneficial to break them up into smaller chunks anyway. May test out this idea by breaking apart the NCBI taxonomy into a set of smaller subsections (bacteria people probably won't need to use reptilian species tags too often..)