feat: add Select() step

[jitingxu1]

Add a Select step, opposite to Drop step.

Use it as column filters in the last-mile preprocessing.

import ibis_ml as ml

recipe = ml.Recipe(
    # to be added
    ml.Select(ml.numeric()),
    ml.ImputeMean(ml.numeric()),
)

Any thought?

[deepyaman]

Sure, makes sense if we have Drop().