tabular data preprocessing improvements

[ahzero7d1]

Additional preprocessing can be implemented for tabular dataset:

1. Add feature standardization
2. One-hot encoding for categorical features
3. Label encoding for multi-class classification tasks

[JulienVig]

We can also check these things at the data connection stage and raise a warning if data is not already standardized/one-hot encoded informing the users that we're doing some default preprocessing