[API] Split /values endpoint in two simpler endpoints

[ForcamPB]

Currently the /values endpont is very complex and could be split in two separate endpoints. One endpoint for querying values. And one endpoint for aggregating values.

[kenwenzel]

I am a bit unsure if the splitting really helps clients to work with the API. I could imagine that it lowers the burden to implement the values endpoint when it is not required to integrate any aggregation operations.

[ForcamPB]

As you said I also see potential in the split in lowering the burden to implement only one of them. And if the two concerns are separated in their own endpoints it's easier to evolve them independently.

[kenwenzel]

One option could be:

• /values (without op, interval)
• /aggregates (with op, interval)

I currently see a small issue with this approach: The Grafana data source uses /values and the "avg" operation to resample the data. Hence the data source would be unusable with endpoints that don't support aggregates.
Do we need some kind of profiles (operation-wise) for the endpoints? Or some kind of discovery mechanism for the supported ops?