More control over shards and indices in elasticsearch

### Problem description

Due to a space problem on one of my elasticsearch nodes the newly created next deflector index could not be allocated and the elasticsearch cluster went in status "red".

I had to use the elasticsearch api endpoints _cat/indices and _cat/shards to find the index/shards that elasticsearch was not able to assign and -XDELETE it. Plus, before deleting some older indices from the graylog server webinterface.

As graylog fully encapsulates the elasticsearch cluster, I wish I had a feature in the graylog webinterface to do such kind of work. Something like the marvel or kopf plugins for elasticsearch.
### Steps to reproduce the problem
1. write your cluster full and then get the deflector index cycled so unallocated shards will render the elasticsearch cluster red.
### Environment
- Graylog Version: 2.0.0
- Elasticsearch Version: latest
- MongoDB Version: latest
- Operating System: CentOS 7.1