Allow more granular clear-sky model rescale for CSI filtering

[kandersolar]

In the clear-sky degradation analysis workflow, the irradiance_rescale step rescales the entire timeseries at once. It would be nice to be able to rescale in shorter chunks, e.g. quarterly or monthly, so that clear-sky model error seasonality doesn't get propagated into the normalized energy calculation.

One complication is that we probably don't want to do chunked rescaling for the irradiance used for normalization -- a big motivation of the clear-sky workflow is to detect sensor drift and chunked rescaling would copy that drift into the CS signal. So I think we would need to have two distinct irradiance signals -- one used for the normalization, and one used for the CS filter.