DM-41118: Add metrics from diffim task metadata

[isullivan]

No description provided.

[natelust]

There are a few suggestions to change things up, what path you choose is up to you.

[natelust]

There is a lot of duplication here, I'm going to make some suggestions that make use of Analysis_Tools designed configurability.

Start by deleting all of the metrics below and replace with one tool:

# Pick whatever name you want
class DiffimMetadataMetricTool(AnalysisTool):
    """This tool is designed to calculate the median of data produced from diffim metric data"""
    # This is a convenience field to make it easier to set things from the "top level"
    # you could also just reach right in and set things in config as well and leave
    # this out.
    metricKey = Field[str](doc="The metadata field to process, name will be used as the metric name")
    units = Field[str](doc="The units to associate with the metric")

    def setDefaults(self):
        self.process.calculateActions.diffimMedian = MedianAction(vectorKey="")
        self.produce.metric.units = {"diffimMedian": "ct"}

    def finalize(self):
        # This function is implemented to support the convenience fields defined above
        self.process.calculateActions.diffimMedian.vectorKey = self.metricKey
        self.produce.metric.units["diffimMedian"] = self.units
        self.produce.metric.newNames = {"diffimMedian": self.metricKey}
   

Alternatively, we could build a tool that takes a configuration of a dict field for all the metrics to produce with one tool

# Pick whatever name you want
class DiffimMetadataMetricTool(AnalysisTool):
    """This tool is designed to calculate the median of data produced from diffim metric data"""
    metrics = DictField[str, str](doc="The metrics to calculate from the task metadata and their units")

    def finalize(self):
        for name, unit in self.metrics.items():
            setattr(self.process.calculateActions, name, MedianAction(vectorKey=name))
        self.produce.metric.units = dict(self.metrics.items())

In this case the yaml would change to look something like:

atools.diffimMetadataMetric: DiffimMetaDatametricTool
atools.diffimMetadatametric.metrics:
  nMergedDiaSources: ct
  nGoodPixels: ct
  ...

This would result in a single atools bundle uploaded to sasquatch with multiple measurements inside it. Both are perfectly reasonable ways to do things, and the dashboards can handle plotting from each the same way. The big difference is that if you are getting these back from the butler, the former will result in n dataset types and n butler.get calls. The later will be a single dataset type and a single butler.get call with multiple values in it.

[natelust]

atools.nUnmergedDiaSources: DiffimMetadataMetricTool
atools.nUnmergedDiaSources.metricKey: nUnmergedDiaSources
atools.nUnmergedDiaSources.units: ct

[natelust]

atools.nUnmergedDiaSources: DiffimMetadataMetricTool
atools.nUnmergedDiaSources.metricKey: nMergedDiaSources
atools.nUnmergedDiaSources.units: ct

[natelust]

atools.nUnmergedDiaSources: DiffimMetadataMetricTool
atools.nUnmergedDiaSources.metricKey: nGoodPixel
atools.nUnmergedDiaSources.units: ct

[natelust]

similar to above for each of the tools below

[natelust]

Probably best practice to import your module here instead of star

[natelust]

The changes below will make sense after reading the metric definition comments.

[natelust]

I'm confused on why these templates are here?