Combining mgpu and mqpu platforms to work in tandem  #911

Required prerequisites

• Search the issue tracker to check if your feature has already been mentioned or rejected in other issues.

Describe the feature

mgpu target allows one to scale qubit count by pooling gpu memory together. It is purely designed for simulation mode.

mqpu allows one to parallelize over multiple qpus. For example, to perform gradient batching, one has to set the mqpu target and use observe_async

I can imagine workflows where we would like to do 40 qubit simulations and batch gradient calculations. This requires the use of mgpu and mqpu platforms simultaneously.

From a usage perspective, mqpu is accessed via the async functionality explicitly and hence may not require one to set the target if we can offload its functionality to async. We can then use mgpu to scale qubit counts whilst having access to mqpu via async.

Thanks team.