Add functionality to use single precision floating number

[hqwang815]

Adding a function to convert all numpy array and constant floating numbers to single precision

[apalha]

I just saw this. You want it to run on GPUs, right? I propose adding an option flag either in the class AcousticSimulation or a module wide variable the user can set. Once this is set all arrays are of a specific type.

I was having difficulties with this, but I think we can just add a variable in the module and whenever we create an array we use the array type.

We can discuss this today also.