Fixed a typo on ML-Agents-Overview.md (#4130)

Fixed redundant "to" word from the sentence since it is probably a typo in document.

docs/ML-Agents-Overview.md

 training on easier tasks can provide a scaffolding for harder tasks in the
 future.
 
-Imagine training the medic to to scale a wall to arrive at a wounded team
+Imagine training the medic to scale a wall to arrive at a wounded team
 member. The starting point when training a medic to accomplish this task will be
 a random policy. That starting policy will have the medic running in circles,
 and will likely never, or very rarely scale the wall properly to revive their