Pl/remove distribution table

[paulflang]

This introduces a couple of changes disussed in the last meeting:

• split up parameter table in a parameter table and random effects table
• Allow multiple random effects per expression
• rename the groupings (I used distributions now)

Additionally, I removed the individuals table. The information there is now implied in the measurements table, via the distributionId columns. These column can be used to express which datapoint belongs to which patient. Or if there are multiple groupings, like patient, tumorStage and eye color, then they define which patient has which tumorStage and which eye color at which time point. One and the same patient has to share the same covariates drawn from the patientId distribution. Ditto for tumorStage and eye color.

I also removed the distributionType column from the parameter table and put it into the YAML file.

[paulflang]

@eraimundez , what do you think in general?
And what do you think about having the random effect table listed at the same level of hierarchy as the parameter table in the YAML file?