This adds the capability to drop metrics matching a regex and labels
that match both a metrics and labels regex.

This is useful for dropping data that is shipped by collectors or
remote write clients that is not desirabel to store.

This adds the same capability as ingestor to collector to drop
labels and metrics matching regexes.  This is useful to drop data
that is not desirable to store from even being collected.  Doing
this as collection time can reduce resource usage on collectors and
reduce the volume of data that ingestors and Kusto needs to process.

This adds a basic HyperLogLog-based cardinalty estimator and a
rolling version that allows for cardinality counting over time.