The boards arrived when I was travelling, and now that I am home again, I can actually work on this again. As a start, I assembled one of them:

I am still waiting for the battery holders — in the mean time, I used one that I had available, which has the + and - reversed — that's why it's soldered wrong. That's actually convenient, because it gives me access to the programming and reset pins under the matrix. Of course the final version will have the batteries under the device.

The USB plug works, but I think I made the pads a bit too wide and the spacing too large — they have a good contact, but it's lost when you wiggle the device.

I have the CircuitPython firmware with the custom pin definitions and increased internal disk size flashed. The next step is to write the matrix-driving code in Python, to test it, and then to rewrite it in C as part of the firmware, so that it can run in the background. I was hoping to start on that today, but there was no time this evening. Tomorrow is a holiday, so I hope to get that working then.