Include uncertainty on other vehicle actions

[huevosabio]

Basically, assume that "obstacles" draw from a distribution over their actions and act accordingly.

This will likely explode the state space, so we should try to come up with a cool method to solve it :)

[huevosabio]

I've implemented random actions drawn from an arbitrary distribution for the obstacles;

However, calcReward does not take into account this new reality.

The issue is that calcReward uses extends the current state using the current action to see if there would be a crash... a more reliable source of truth would be to calculate reward based on the original and following state

[huevosabio]

Ahh actually, I think we would be better by keeping it as is...

[huevosabio]

In MCTS, we simply simulate soooo we can actually use simulated obstacle actions to see what our expected value will be. In FS, an option would be to simply get the expected value.