also, verify all 3 first bytes of SPI commands. I'm not sure why
I wasn't doing that, probably because I was getting a lot of AVR
err and thought that only 2 bytes of the cmd were echoed. But now,
with a reliable SPI setup, verifying 3 bytes seems to work.

After many trials and errors in reliably accessing AVR chips through
my SPI relay design, I resigned myself to accepting 125kHz communication
speed with it. I find the complexity of solutions allowing to keep 250kHz
speeds to be excessive.

Also, remove leftovers from previous move.

The batch mode thing seems to have been caused by bad timing on my
prototype. Now that I've corrected it, live interaction seems to
work fine.