Explore constraints

Comment from @ypar in https://github.com/greenelab/Batch_effects_simulation/pull/6#issuecomment-529592611

```Also consider proportions or minimal samples beyond the absolute count of experiments. i.e. was it important that you had 20+ experiments or at least 21 experiments that contained Z samples, which was Y proportion of the total and represented X % variation? was it not important? would it hold true if you had 21 * 1000 samples each experiments vs 21 * 1 sample each experiments?```