TNRS times out on large trees

[chinchliff]

It looks like the taxomachine neo4j on the dev server has a maximum connection timeout of about 3 minutes. I am not sure how to increase this, but it definitely needs to be much longer if we want to get TNRS results for large trees.

An alternative would be to use a local instance of taxomachine to do the TNRS. This would require downloading the taxonomy. We could use/provide a snapshot of the taxonomy db so people would not have to build it. This is related to the issue of bringing taxonomy into OTU. See #14.

[chinchliff]

An update, the version of OTT that we import into OTU is bare-bones. It will not contain the indexes required to do local TNRS. So, to facilitate that, we will need to run another neo4j server with the taxomachine plugin and a taxonomy database. We should:

1. Make a snapshot of the database available
2. Write a script (python) to be called via the webpage that will:
   a) download and configure a neo4j instance to run the taxonomy server alongside the otu neo4j
   b) download and unpack the snapshot and install it into the taxonomy neo4j
   c) set properties in the otu graph to indicate that local tnrs is available