remove FAR string tags which break aggregation

A bug was found in the test suite which caused a slight over-counting in some of the triggers. Basically, we rely on the data being aggregated by min FAR, so that for each injection we only keep the trigger in InfluxDB that was the "best". In influx, if you write a measurement with the same time and same set of tags, the first one will be replaced by the later. However, if the tags are unique both data points will be kept. Therefore, since we used tags to store the FAR value as a string, this led to some duplicated points, ie a trigger was stored for an injection with FAR 2/day and later a trigger was found for the same injection with FAR 2/year.

The effect is a small (<5%) over-counting in the number of "found" and "insignificant" triggers. Fixing this leaves the efficiency metrics unchanged or slightly increased.

Since we've now transitioned to using trigger-type data which arbitrary number of numeric columns, rather than time-series type data with a single numeric column, we no longer need the FAR to be included as metadata in the tags since we have the numeric FAR value attached to all triggers. This MR just removes the obsolete (and problematic) FAR tags.

See an example of a duplicated trigger on a test suite dashboard using old version of code

And the same trigger on a test suite dashboard using this version of code