Memory management strategy of temporalite

What is the memory management strategy of temporalite in-memory mode?
Is there an upper limit of the memory consumption?
I observed that the memory consumption continuously growing as number of workflow executions grow:

If there is an upper bound, then:

  1. How can we configure it?
  2. What happens when it is hit?

Temporalite is similar to auto-setup as in it starts all service roles in the a single process. This makes it imo hard to find out which service would need to be looked at (especially without server metrics).
My best guess would be to try to look at dynamic config settings for history service cache, see more info here.