My only concern, skimming that, is that the digispark and micronucleus boards want to run everything at 5V, rather than 3.3V. Is that going to be an issue for 3.3V I2C devices further down the line?
Other than that, it isn’t so hard to make a digispark based device, with a USB socket (rather than male tabs) - for the suggested “long USB cable, short I2C cable” strategy, and then a 3.5mm stereo jack at the other end conforming to the tip-SDA, ring-SCL spec I seem to be perpetuating).