Suggestion to refactor optuna code

[sfmig]

This PR builds on work from #136.

Contributions:

• reduces code duplication,
• makes the hyperparameters to sweep across optional. Before, it only learning_rate or n_epochs were passed it would throw an error,
• ~makes the hyperparameters we can sweep across more flexible.~ --> removed since it was potentially less clear
  • ~I am happy to keep in the guide that we only consider n_epochs and learning_rate for now, and we could just test those, but maybe this approach makes it easier to implement other in the future.~

It may be a good idea to integrate the tests from #136 here.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 25.53191% with 35 lines in your changes missing coverage. Please review.

Project coverage is 36.76%. Comparing base (4a705b8) to head (6a84a16).

Files	Patch %	Lines
crabs/detection_tracking/train_model.py	28.00%	18 Missing :warning:
crabs/detection_tracking/optuna.py	25.00%	12 Missing :warning:
crabs/detection_tracking/models.py	16.66%	5 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #188 +/- ## ========================================== - Coverage 37.17% 36.76% -0.42% ========================================== Files 19 20 +1 Lines 1310 1352 +42 ========================================== + Hits 487 497 +10 - Misses 823 855 +32 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.