395 implement new deterministic metrics kge

[samland1116]

This PR updates the metrics framework for KGE, KGEmod1, and KGEmod2 to allow for user defined constants. This PR also explores the impact of wrapper functions on query performance. By creating a wrapper around the KGE functions that ingests the KGE base models, we can specify user defined constants like below:

from teehr import DeterministicMetrics

kge = DeterministicMetrics.KlingGuptaEfficiency()
kge.sr = 0.5
kge.sa = 0.5
kge.sb = 0.5

include_metrics = [kge]

metrics_df = ev.metrics.query(
    group_by=["primary_location_id", "configuration_name"],
    include_metrics=include_metrics
).to_pandas()

Alternatively, the existing method to query metrics still works despite the presence of a wrapper, shown below:

from teehr import DeterministicMetrics

metrics_df = ev.metrics.query(
    group_by=["primary_location_id", "configuration_name"],
    include_metrics=[DeterministicMetrics.KlingGuptaEfficiency()]
).to_pandas()

Unexpectedly, the presence of a wrapper that does not alter values within a given function results in somewhat significant performance improvements while the presence of a wrapper that does alter values within a given function results in performance similar to that of our existing metrics implementation.

I propose we implement this functionality on all existing metrics functions to standardize metrics behavior while simultaneously improving metric query performance.

[samlamont]

I think it looks go to me overall. I wonder if we should just implement the wrappers for all metrics in this PR since it would get rid of the requires_wrapper checks and allow us to call the pandas_udf in the same way for all metrics?