selectable aggregate list should be explicitly de-duped

Repeated aggregate functions in a single statistics query are problematic.

For example, requesting count twice:

  GET /v2/meters/cpu_util/statistics?aggregate.func=count&aggregate.func=count

results in the count reported via the mongo driver being double the real value.

Worse, multiple references to cardinality results in a Nan being reported for the additional aggregate:

  GET /v2/meters/cpu_util/statistics?aggregate.func=cardinality&aggregate.param=project_id&aggregate.func=cardinality&aggregate.param=resource_id

   ... "aggregate": {"cardinality/project_id": 1.0, "cardinality/resource_id": NaN}

Instead the aggregate list should be de-deduped in the API layer, as is the groupby list currently.