separate predictive learning loops from attentional ones

[rcoreilly]

The demands of each are not compatible, and there are two separate pulvinar maps of visual areas, in addition to possible core vs. matrix distinction. Need to do more research, but computationally need to go back to having many layers contribute to each pulvinar prediction, especially lowest level -- that was a key feature of cemer version that has been lost.. but that is not compatible with focal attentional pool-organized topo loops that may work best as more closed loop rather than as dramatically open-loop..

[rcoreilly]

This has been a very productive fix -- now incorporated into all models.