Eemeter

-----


* Force index to use nanosecond precision
* Compute coverage using same offset as initial reads to fix issues when downsampling hourly data
* Update test data location
* Fix bug in daily plotting to remove NaN values if input
* Refactor sufficiency criteria to be more explicit and easier to manage

-----

* Correct dataframe input behavior and final row temperature aggregation
* Remove unnecessary datetime normalization in order to respect hour of day
* Convert timestamps in certain warnings to strings to allow serialization
* Allow configuration of segment_type in HourlyModel wrapper

-----

* Update daily model methods, API, and serialization
* Provide new API for hourly model to match daily syntax and prepare for future additions
* Add baseline and reporting dataclasses to support compliant initialization of meter and temperature data

-----

* Addition of modules and amendments in support of international facility for EEMeter, including principally:
* Addition of quickstart.py; updating setup.py and __init__/py accordingly.
* Inclusion of temperature conversion amendments to design_matrices; features; and derivatives.
* Addition of new tests and samples.
* Amendments to tutorial.ipynb.
* Addition of eemeter international.ipynb.
* Change .iteritems() to .items() in accordance with pandas>=2.0.0
* .get_loc(x, method=...) to .get_indexer([x],method=...)[0] in accordance with pandas>=2.0.0
* Updated mean() to mean(numeric_only=True) in accordance to pandas>=2.0.0
* Updated tests to work with pandas>=2.0.0
* Update python version in Dockerfile.
* Update other dependencies (including adding rust) in Dockerfile.
* Remove pinned dependencies in Pipfile.
* Relock Pipfile (and do so inside of the docker image).
* Update pytests to account for changes in newer pandas where categorical variables are no longer included in `df.sum().sum()`.
* Clarify the functioning of start, end and max_days parameters to `get_reporting_data()` and `get_baseline_data()`.

-----

* Update observed_mean calculation to account for solar (negative usage) to provide
sensible cvrmse calculations.

-----

* Remove missing hour_of_week categories in the CalTrack hourly methods so they predict null for those hours.