Rework the classes.py for scaling and performance

[bjoernenders]

classes.py is our main module for holding data in PtyPy. Currently we are facing performance issues when scaling up to millions of views. Also the dynamic resizing nature of the data holding numpy arrays can cause trouble further down the road when using GPU acceleration. I'm making this issue to remind myself what work needs to be done to address performance and scaling issues:

Here are a few things to look at:

• Make View and POD class as thin as possible.
• Create memory buffer/numpy to hold all relevant data for views to be created.
• Create similar buffer for POD
• Allow forming of POD and View instances from buffer arrays
• Consider a fixed canvas for Storage classes and make sure the view access doesn't make errors
• Consider tiling for shared Storages
• Make sure methods modular enough for easy subclassing to GPU memory.