This is a very useful feature for many of the dashboards,
when not using the merge flag. These dashboards begin
with a metric-list, and then do a statistics call for each
unique set of dimensions returned. For many dashboards,
metrics are returned where no data is currently being
collected (deleted VMs, etc), causing many unnecessary queries
for no data -- not to mention an ugly dashboard with tons
of noise.
This enhancement will help both the current grafana implementation
as well as our 2.0 port. Note that this patch supports the java
implementations for vertica and influxdb, as well as the python
implementation of the api for influxdb (vertica not currently
supported in python).
Change-Id: I683f2a53aaf2d2ad8005dd1542883636086aad4a
bc629f612f