Actually this is a good question - there's a lot of control engineering research on Data-Driven Control Systems.

Check Steve Brunton youtube channel, he is one of the leaders in this area: https://youtu.be/gb_C9LcjDSI?si=xUjqUZ9-0MIFohX6

If I had to bet, I would bet against it. Boston Dynamics for example, for the longest time, didn't use anything other than Model-Predictive Control. Only recently have they started using RL

No. Way too hard to validate.