/controller/startGracefulFailover end point accepts multiple nodes when it should not

Problem

Currently graceful failover can only be done on one node at a time unlike hard failover which support multiple nodes.

Unfortunately /controller/startGracefulFailover does accept multiple nodes being passed in. It will do a graceful failover of the 1st node and ignore the others.

Steps to reproduce

1. Setup a 3 node cluster with 1 bucket and 2 replicas
2. Do a graceful failover of two nodes:

   curl -XPOST --auth Administrator:password http://10.0.0.1:8091/controller/startGracefulFailover -d optnode=[10.0.0.2:8091,10.0.0.3,10.0.0.3.8091]

Expectation
If multiple nodes are passed in an error should be returned.