Create a higher-level abstraction for managing GPU memory

[kjkm]

REQUIREMENTS
This probably needs to do the following:

• Replace manually defining buffers and bind groups any time we want a new chunk of memory
• Nicely organize memory across multiple bind grounds (not just bind group 0, as we are currently doing)
• Should replace existing memory allocation functions (like createBindGroupLayout)
• Should employ a more object-oriented style (a single class, which contains all of the functions needed for memory management)
• Should be a singleton memory manager
• The memory required for any given compute pass should get encapsulated into a single reference so that it can be easily fed to the function in Create a separate function for handling a single compute pass #46

These reqs are not exhaustive and more work will need to be done to architect this out