Investigate event, baseflow, rising limb filter algorithms

[mgdenno]

Several of the proposed signature metrics we have identified as important require filtering the timeseries based on identifying some pattern in the timeseries that requires a wholistic view of the timeseries not just checking a single value against some criteria (i.e., a threshold). We need to understand what is currently being done in this area (event, baseflow, rising limb) as well as what may be required in the future such that we can generalize this process to handle not only current needs but future ones too (that should be the aim anyway - in reality it can be hard to anticipate future needs).

[mgdenno]

I think we should investigate the "Series to Series" pandas_udf() for this. See: https://spark.apache.org/docs/3.4.2/api/python/reference/pyspark.sql/api/pyspark.sql.functions.pandas_udf.html

If this works the way I think it does, it could be an easy way to add event detection with existing Python code (from HydroTools). We could then later see if doing it all natively in PySpark would be more performant.