Full codebase redesign

[swillner]

• refactoring of modules
• remove code duplication
• separate technical from scientific modelling
• abstract from statistical modelling library (pymc3)
• use xarray
• various smaller changes, e.g. simplified tests

[rgieseke]

After a previous PR (#101) had a tiny bug we've made sure to re-run the full test suite to ensure all variables are having the same results as initially.

=== test session starts ===
platform linux -- Python 3.11.10, pytest-8.3.4, pluggy-1.5.0
rootdir: /home/robert/Projects/attrici/attrici-private
configfile: pyproject.toml
plugins: anyio-4.6.2.post1
collected 42 items                                                                                                                                                                                                                   

tests/test_detrend.py .........                                                                                                                                                                                                [ 21%]
tests/test_preprocessing.py ..                                                                                                                                                                                                 [ 26%]
tests/test_vendored_ssa.py ..............................                                                                                                                                                                      [ 97%]
tests/test_version.py .        
==== 42 passed ===

[rgieseke]

We've found a tiny bug for logp reporting and fixed it with two additional commits. Note that for the variables pr and tasrange the values in reference data are different due to the tiny bug fixed in #101 so these two are skipped (values are in the same range though).

[matthiasmengel]

Great progress, thanks!