Make Retention Heatmap visualization inputs dynamic

I like the Retention Heatmap visualization a lot and plan to use it for a number of cohort analytics use cases. It is however quite restricting that a pre-aggregated input for value & cohort size is needed.
If we could use any field and dynamically aggregate, say COUNT_D(user_id), this would allow us to employ any type of filter to change the cohort definition on the fly.

Agreed - having to hard code really restricts the analytical potential of this viz