Add `params_to_tune` for catboost models

[Mr-Geekman]

🚀 Feature Request

Define params_to_tune for catboost models.

Proposal

Sources for the grid:

• pycaret grid for catboost
• optuna example
• lama

Suggested grid:

{
    "learning_rate": FloatDistribution(1e-4, 0.5, log=True),
    "depth": IntDistribution(low=1, high=11, step=1),
    "random_strength": FloatDistribution(low=0, high=0.8),
    "l2_leaf_reg": FloatDistribution(low=0.1, high=200, log=True),
}

Test cases

1. Check that params_to_tune isn't empty for catboost.
2. Check that it is possible to sample a point from a given distribution and set it in the model.
   • Iterate over each parameter
   • Sample value using RandomSampler().sample_independent(None, None, name, distribution).
   • Set new parameter value using set_params
   • Check that value is set

Additional context

No response

[Mr-Geekman]

Closed by #1185.