Different Environment Shape in Training For Reinforcement Learning

Question

I am trying to make an RL model for scheduling. The goal is to find the best spot on a schedule to insert a task with different set of constraints. My environment can be seen as