DJORNL parser improvements

[ialarmedalien]

A few improvements for the parser:

• cluster data could be parsed and merged directly into the node data, rather than uploading all the node data and then doing the cluster data
• check for additional (invalid) headers
• the 'GO description' field is currently not populated correctly (due to problems in the input data); it could either be generated by pulling in the GO term names from the existing RE db (i.e. have a copy of the GO term data in the DJORNL node document), or the search could be to query over the joined DJORNL and GO tables.
• add a 'validate' function to parse the data and validate it, without uploading it.
• add a README in the djornl directory with information specific to the dataset.

[jayrbolton]

Closed by #47 and/or #48 and/or #49 and/or #50?