adjustments to map weighting

[DHekstra]

We switched from using an "alpha" like in rs-booster to using a percentile of the amplitude differences distribution to determine the point where weights kick in. The latter percentile cannot go over 100, meaning that weighting is limited in its maximum extent. We want to be able to do both.

See

• https://github.com/Hekstra-Lab/drug-screening/blob/main/valdo/helper.py#L203
• https://github.com/rs-station/rs-booster/blob/main/rsbooster/diffmaps/weights.py