Is this how it works in a summary?
Writes during active display alternate between X and Y position, depending on the toggle. The X position of the sprite doesn't get updated until the Y position does, because of the way the toggling works. During active display all writes go to both the LoOAM and HiOAM at the same time. Writes to HiOAM are not affected by the toggling. Which sprite it updates depends on which cycle/pixel during active display. During H-blank, writes go to either the HiOAM table of whatever sprite is currently being evaluated, with the backmost sprites being evaluated first, or the frontmost sprite of the last scanline to have sprites on it. I'm guessing each sprite always takes two cycles to evaluate, both during active display when it is determining if a sprite is on scanline or not, and during H-blank when it's doing the VRAM pattern fetches.
The stuff I don't know is, what happens if you write TWICE during h-blank? Will it effect X and Y of the corresponding sprite? Or the name/attributes word of the corresponding sprite, or does that depend on either you're writing to an even or odd cycle? Does DMA always occur on even cycles during active display, or does it occur on odd cycles on odd numbered scanlines?
As a non-programmer, does that mean it is or is not possible to do sprite multiplexing on SNES?
To me, it sounds like you're saying yes but in a very limited way that wouldn't really be useful in most cases anyway. Although, if you could describe some actual potential use-bases in a game then that might help me understand it better.
It sounds like it is possible but in very limited and convoluted. Like, you would have to do several runs at precisely the right timing, and have to watch out for the DRAM refresh messing up the timing, although that can probably be overcome by changing the OAM priority rotation setting. I still don't know if it is possible to change the attributes and name, or just the X and Y coordinates.