mark anomalous metrics as such

[zmarcantel]

I added a bool field to the data.Datapoint struct that will say "yes, I am an anomaly".

This helps applications parse the output channel without creating a second channel that could potentially block if not properly/diligently read from. It also avoids a less-than-ideal strings.HasSuffix("anomalous") check.

I'm open to discussing other methods, but this seemed the easiest.

[mattrco]

Thanks for this. At the moment the three sigma analyzer has a separate output channel for anomalous values; is this insufficient for the use case you're thinking of?

I'm not opposed to merging this, just want to understand the use case better.