Issue: Experiment comparison: Charts are scaled in a way that makes them all look the same

[Philipp-Userlike]

Issue you'd like to raise.

When comparing two experiments, all the charts are scaled in a way that makes it impossible to judge the scale of the difference. They all have the same basic shape.

In the example I would want the 4th chart (role) to visualize that there is a big difference, while e.g. the 5th one is only a very small difference.

Suggestion:

This works well in your default metrics:

For our use case it would be ideal if all charts are scaled this way.