Discuss and document behavior of max_counts at different levels

[ethanwhite]

Due to the complexities of the data we have available max_counts is accessing two different sources of data and using different combinations of sources depending on the level selected. This should be documented with a particular focus on level = "subregion" which actually takes values for different subregions from different sources in a manner likely to confuse users.

This approach may also cause some complexities for our multi-scale work, so we should also consider including an option that allows aggregating colony level data for all levels.