polaris learn: Allow user to use gridsearch during the checking cross dependency process
Currently, hyperparameters are hardcoded for the models used in cross dependency analysis (XCorr)
We do not use any hyperparameter optimization method when making predictions. We should do so by first adding the possibility to use gridsearch when training the model(s).
- propose a gridsearch option to XCorr class
- save the gridsearch resulting hyperparameters per telemetry (e.g. MLFlow logging)
- reuse all found hyperparameters to calculated cross dependencies
NB: for future improvement (out of scope here)
- execute gridsearch only when necessary == when hyperparameters are missing for the telemetry