Hi Phil, thanks for using our products. Right, RTIMU is pretty old, but works pretty well. I haven't found anything more appropriate.
The software can get a lot more complex when reading multiple IMUs at once and recording, so there are a lot of potential causes. Have you timed any particular sections of code to determine what's slow? It could be some 'simple' stuff like I2C clock, non-optimized SD card writes, etc.
I believe the Wireling library has a delay in it that is not totally necessary, and could cause a significant slowdown in this type of application.
With the amount of math RTIMU does, I believe it can run the algorithm at around 250Hz with the default settings on TinyZero. So I think 50Hz or better is achievable for four IMUs as long as the SD writing is optimized. This assumes that the fusion algorithm needs to run on the microcontroller.
Thanks,
Ben