[WIP] Update of prediction intervals

[snath-xoc]

This PR updates the HGBRQuantile and BinnedQuantile Regressor based predictions using the updated skrub data operations. Tutorial helpers are also updated so as to read in the formatted data. ping @jeromedockes @ogrisel

To do:

• Undo the skip of prediction_interval.py in the pixi.toml
• Fix some bugs in the bqr pipeline
• Add in tabicl based quantile predictions

[snath-xoc]

@jeromedockes I added in TabICL, it does look like it performs better (see reliability diagrams for lower and upper quantiles). I am not sure whether we should just merge it to the prediction_interval.py section or keep it separately as an independent file, what do you think?

Tutotial helper for plotting needed to be changed a bit as well due to data type mismatch but otherwise not much more changes there.

[jeromedockes]

thanks @snath-xoc !

I checked that both notebooks are running and giving the expected results, so I think it can be a good idea to merge it so that all the main pieces are in there and then do further improvements with smaller iterations

[snath-xoc]

Alright, just updated the plot_predictions @jeromedockes, @ogrisel and @glemaitre we can merge and then add in more useful explanations/iterate?

[ogrisel]

Merging and let's iterate.