Thanks Igor for looking into this, your assestment is correct. We found out that we had a faulty 3DR I2C splitter. This also explains why the Baro also went bad and we were not able to disarm the motors either since they both are on the same I2C bus.
You also pointed out that the Navio2 (which we are not testing with it yet) now only uses the I2C bus for baro and accessories; which is good since we at least we’ll be able control the PWMs if this happens again. However, I’m still concern that we have 2 accessories connected to the I2C bus and they can make the Baro go bad if they shutdown the I2C bus. Can we use an external laser altimeter like the SF11 connected via UART as the EKF_ALT_SOURCE and take the built-in barometer out of the I2C equation?