Investigate how to handle posting batch readings

[jessetham]

Currently our API only handles creating one reading at a time. This should probably be improved to allow for batch creations since that would decrease the amount of HTTP requests that the sensor would have to do.

One possible implementation is detailed here. Instead of creating a new resource location though we can just return the locations of each newly created reading in the body. The /batches endpoint should be enough to self-document the difference between whether or not the Location header will be there or not.