Implementation of `MPS` support for the simulations

[talagayev]

Describe the feature you would like to have

It could be interesting to adjust the protocol shown in the tutorial down below. This would increase speed for the simulations on certain GPUs.

Would be more interesting for the automatization side

Additional context

There is a new blog post showing the use of MPS in OpenMM:
Here