Size of function_id within Action Space

[hagopi1611]

In mentalgym.py -> self.action_space:

We can leave function_id spaces.Discrete(self.number_functions) as is,

where number_functions should be an upper bound on the total number of atomic + composed functions that we can have in our function bank. We only need to add a persistent variable in the agent's scope that lets it know the total number of functions available to it, which will be the current number of functions in the function bank. The agent will then restrict his action to select only from available functions.

In other words, we don't need to grow the size of the number of actions the agent can take over time.

[LunarEngineer]

@hagopi1611 I think this has been resolved for the moment; I'm going to close this and we can revisit at a later date if our state representation changes.