I think HN would be well served to have more front page retro hardware hacking articles. They may not be useful in the same way as other posts, but they sure as heck are delicious brain food and a relaxing diversion.
The command set looks like a modified HPGL.
It seems pretty accurate, to within a couple of millimeters.
The brown rubber wheels seem to be the only things in contact with the paper.
There were also some plotters with spiky steel wheels at the paper edges to feed the paper, those would work a bit better when new but the spikes would wear off fairly quickly.
Edit, just remembered: it was probably to output one of the VLSI CAD layer formats, like GDS or EDIF.
We also managed to put one of those HP's through its design life in less than 2 years because of how intensely we used it.
Edit: And on second glance it's quite possible that 410's Z80 is clocked faster than the 6502 in the Apple II.
However, I spot a 4.9152 MHz crystal on the photo of the PCB (a quarter from the left, halfway down)
So, I guess it either is overclocked or (IMHO more likely) that that clock is divided into a 2.5MHz clock.
Either way, I agree with you that it likely had a faster clock than the 1MHz of an Apple II.
However (certainly if you can get around with 8 bit registers; it shifts a bit if you can use the 16 bit parts of a Z80) a Z80 has to be clocked faster (some people claim by a factor of about 2:1) than a 6502 to be faster (extreme example: a NOP takes 4 cycles on a Z80, 1 on a 6502)
In summary, I guess this _is_ faster than the Apple II it was designed to connect to, but not by much.
Even the little Commodore 1520 was a very capable machine in a toy form factor. Four independent color pens, and a great tool for learning geometry and programming at the same time.
To digitize it I covered the wall of a spare office with graph paper and projected the film onto the graph paper and manually using pencil marked up the graph paper then transcribed the x,y coordinates which I than plotted on a HP A3 plotter.