Refactor GCMD harvester

[hmaier-fws]

From PR #12:

The GCMD harvester was written from scratch using a completely different process. It now predominantly uses the concepts/ endpoint to gather the data and build the tree. It's still recursive, but now, for each keyword at least 1 request is made to the GCMD API to harvest the data and build the tree. It modifies both harvesters/src/gcmd.js and harvesters/conf/gcmd.json.

It creates the following structure:

{
  "uuid": "",
  "label": "",
  "broader": "",
  "definition": "",
  "children": [ ]
}

Note: if no definition found then "No definition available." is used instead.

New harvester control process to build all GCMD vocabularies

New harvesters/src/gcmd-all.js

Along with the new harvester process is this control process that will build all the GCMD vocabularies listed in the config file harvesters/conf/gcmd-vocabularies.json. It also uses harvesters/conf/gcmd-all.json config file and generates resources/gcmd-vocabularies-dynamic.json.