Add default params_to_tune for missing values transforms

[Mr-Geekman]

🚀 Feature Request

Add default params_to_tune for missing values transforms.

Proposal

Suggested grid for ResampleWithDistributionTransform: None. Because it isn't used for improving quality.

Suggested grid for TimeSeriesImputerTransform
if seasonality == 1:

{
    "strategy": CategoricalDistribution(["constant", "mean", "running_mean", "forward_fill"]),
    "window": IntUniformDistribution(low=1, high=20)
}

if seasonality > 1:

{
    "strategy": CategoricalDistribution(["constant", "mean", "running_mean", "forward_fill", "seasonal"]),
    "window": IntUniformDistribution(low=1, high=20)
}

To discuss:

• Can we somehow add "seasonal" to strategy? It requires some knowledge about the frequency. We can also tune window in this case.

Test cases

Look at #1221. Check empty grid as well.

Additional context

No response

[Mr-Geekman]

Closed by #1232.