It was meant mostly for easy access to the bus. SPI handles IMU and GPS and is pretty loaded with traffic, I would not suggest connecting anything additional there.
Developing an autopilot for Raspberry Pi is a lot about bandwidth management. It only has 2xSPI (1xSPI in official HAT), 1xI2C and 1xUART. You would want to have UART on DF13 for telemetry, so we can’t use it internally. We have to move data from 2xIMU, 1xGNSS, 1xBaro and IO co-processor with minimal latencies and at very high rates. To optimize flight performance we have occupied all buses available on the 40-pin connector and this is why we suggest to use USB for extensions.
You can try. What I would suggest to do is to do some bench testing and monitor parameter group “PM” in the logs. Make sure that there are no spikes in MaxT over 3ms and that Nlon is not growing significantly. These are main indicators of performance.
I would recommend replacing the FLIR Lepton breakout board with PureThetmal 1 breakout. It provides plug and play connectivity over USB at up to 9Hz. I too would recommend the cameras from Seek Thermal, however they have not been forthcoming with their promised SDK, so integration will require some reading.