331 feat adding spearman pitch control

[twar-TuVie]

First implementation of pitch control model after Spearman(2017)

[codecov]

Codecov Report

❌ Patch coverage is 13.04348% with 100 lines in your changes missing coverage. Please review.
✅ Project coverage is 97.41%. Comparing base (e4ce2b3) to head (3156ba4).
⚠️ Report is 79 commits behind head on develop.

Files with missing lines	Patch %	Lines
databallpy/features/pitch_control_spearman.py	11.11%	88 Missing ⚠️
databallpy/schemas/tracking_data.py	25.00%	12 Missing ⚠️

❌ Your patch check has failed because the patch coverage (13.04%) is below the target coverage (90.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop     #362      +/-   ##
===========================================
- Coverage    99.22%   97.41%   -1.82%     
===========================================
  Files           49       67      +18     
  Lines         3736     5291    +1555     
===========================================
+ Hits          3707     5154    +1447     
- Misses          29      137     +108     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Alek050]

Hi @twar-TuVie,

Thanks for opening the first implementation of the pitch control!

I have a few points that relate to performance for now:

1. build_ndarray and prepare_data together call multiple regex matches per frame. Regex is relatively slow. I would suggest pre-building arrays per frame in tracking_data.get_spearman_pitch_control instead. This way, you can also use vectorized computation to do it over all frames at ones instead of re-computing for every frame.
2. in get_spearman_pitch_control_single_frame try to pre-create empty np arrays using np.empty since appending over large lists will become slow. Also change default n_x_bins and n_y_bins to (32, 24) or something, that will be good enough for 95% of use cases and make it 4 times quicker in computation.

These are just some first observations. A few questions to make this more productive:

1. Is there anything else you want feedback on?
2. I will go more in depth in optimization of the code later on, did you try a line-profiler yet to see what the current bottlenecks are?
3. Can you point out which parameters were not present in the original spearman paper? (e.g. max_velociyt and reaction time?)

The CI test are failing because the code is not formatted the way it should be. runn the following commands to fix the linters:

poetry run ruff format tests/ databallpy/
poetry run ruff check --fix tests/ databallpy/