DMA code generating

I've seen discussion about generating code to speed up vram writes on the NES and so I'll might as do the same with the SNES's DMA to cut down on the CPU overhead between DMA transfers.

I'm also planning on using tricks such placing the stack on the $43xx registers and pushing some "effective addresses" onto it. Somebody once mentioned that it's possible to have an extra DMA channel for vram addresses, and it looks like it will benefit here.

Here's how the generated code will look like:



If you find anything wrong with the code, please mention it.