add data profiling to features and raw data

This is part of the "common" pipeline which is more bespoke than the main pipeline. The question is: should this be treated as an automated feature, or a custom add on?

New idea for turning features like this on or off: collect these rules into a separate footer which is conditionally included based on the value of a variable in the task-specific makefile. So here, we could have something like:

PROFILE_RAW_DATA := FALSE
PROFILE_FEAUTRES := TRUE

Downstream, we could have stuff like:

REGRESSION_TASK := FALSE
CLASSIFICATION_TASK := TRUE
PERMUTATION_OK := TRUE

to switch groups of evaluation metrics on or off. This is a whole other thing, need to create a new issue.