Rework validation for measurements and pending_experiments

[Scienfitz]

Fixes #453

Notes re measurements validation

• Two utilities (one for parameters and one for targets) for input validation has been extracted. Additional validations for binary targets have been added. The utilities contain parts from add_measurements and fuzzy_row_match
• As a result fuzzy_row_match does not perform any validation anymore. Whether numerical parameter measurements outside of numerical tolerances are allowed is now decided by outside of this function, it will always match numerical parts on the smallest distance
• add_measurements now simply calls the utilities
• tests for invalid parameter input have been extended
• Any duplicated validation is excluded due to validated dataframes being wrapped in ValidatedDataFrame

Notes re pending_experiments validation

• Campaign.recommend now performs validation of pending_experiments
• tests for invalid pending_experiments have been added

Notes on recommenders

• Pure recommenders now validate measurements and pending_experiments in recommend
• Despite being derived from pure recommenders, bayesian recommenders have their own implementation of measurements and pending_experiments because the validation needs to happen before the call to the base class
• meta recommenders dont perform any validation as at the base level there will always be a pure recommender which performs validation if still needed

Notes on numerical_measurements_must_be_within_tolerance

• The only option to activate this behavior is for measurements in Campaign.add_measurements
• In particular, i) pending_experiments are always valid independent of any numerical tolerance deviation (both in recommenders and campaign) and ii) measurements are always valid independent of numerical tolerances for recommenders

[AdrianSosic]

Hi @Scienfitz, thanks the refactor 🏗️ Below my comments

[AVHopp]

LGTM

[AdrianSosic]

First round of very minor comments, you'll still get a full review

[AdrianSosic]

So now here's finally your full review 🙃