Eemeter

-----

* Add GHI sufficiency check requiring 90% coverage for each month
* Add weights propogation from data class to daily model via "weights" column
* Converted daily model settings from attrs to pydantic
* Refactored daily model initial guess optimization to use consolidated optimize function
* Add experimental daily weighting for hourly model fitting (if one day is crazy, it will be down weighted in the fit)

-----

* Add new hourly model to support solar meters and improve nonsolar results

-----

* Add github action to publish to pypi
* Bump to latest packages and remove all deprecation/future warnings as of 2024-12-20.
* Allow identical observations to not raise exception for daily model in `linear_fit`.
* Handle ambiguous and nonexistent local times when creating billing dataclass
* Fix serialization and deserialization of hourly CalTRACK metrics.
* Rename HourlyBaselineData.sufficiency_warnings -> HourlyBaselineData.warnings
* Add disqualification field to HourlyBaselineData and HourlyReportingData
* Fix bug where HourlyBaselineData and HourlyReportingData wasn't actually NaNning zero rows when `is_electricity=True`.
* Constrain eemeter daily model balance points to T_min_seg and T_max_seg rather than T_min and T_max.
* Fix bug in `linear_fit` due to SciPy's `theilslopes(y, x)` not following the same order as `linregress(x, y)`

-----

* Handle ambiguous and nonexistent local times when creating daily dataclass

-----

* Update docs.
* Update typehints on core daily and utility functions.
* Minor change to loading test data to ensure the reporting period is a year ahead of the baseline period.

-----

* Flip slope when deserializing legacy hdd_only models