* AF performance - not yet, but AF performance like a 1D-X in a MILC is just a matter of some smart algorithms and enough CPU ooomph ...
If Canon decides to really commit to mirrorless I think their dual pixel approach is going to make this happen pretty fast. We already know that dual pixel is superior to traditional phase detect for lenses faster than f/2.8.
* battery charge - not equal, partly due to higjher power requirement from EVF/LCD, but in equal parts because manufacturers are sticking freakin' wimpy batteries into MILC bodies that could just as well hold regular DSLR batteries if only the handgrip was made ever so slightly larger
This is my biggest beef with current mirrorless cameras, there's so much of a push to make them as small as possible that they've completely thrown out ergonomics except for maybe the EM-1 and GH4.
* EVF ... in bright daylight not as good yet as a really good EVF ... better than rebel mirror-viewfinders. In really poor light better than any DSLR viewfinder
I strongly disagree with this, low light is where the EVF is weakest. Tracking motion in bright light with an EVF is fair at best, tracking motion in low light with an EVF is a complete no-go. I'd love to see a hybrid where you have a traditional OVF with the option to add an EVF onto the hotshoe as need, but I don't see myself give up the OVF any time in the near future given the current state of EVFs.
That technology may also just as well eliminate the need for heavy and expensive ground/polished glass lenses.
I don't know about that, but I do think as embedded processing power increases and deconvolution algorithms improve you'll start to see some very small and optically simple lenses that produce images comparable to much more complex lens designs. Canon certainly seems to have a large lead in this area with DLO and I suspect they'll try to integrate that into DIGIC at some point.