TODO

 * Test restarting IMU from a wedged state via watchdog
 * Setting yaw center position interacts with phase encoding in an
   annoying way
 * Update counts to be 10ms not 100ms, so that the fire count can be
   more precisely controlled?

 * Rip out all the CubeMX crap and integrate more cleanly

r4 ERRATA
  * rs485 has A/B swapped vs pi3 hat and IMU junction board... :(
  * need a lot more bulk capacitance for the AEG, at least 100uF
    effective at 20V, probably more like 200uF.. either that or get a
    driver that allows a substantially higher PWM frequency
  * mounting holes leave a lot of wiggle room when installed

DONE 2019-05-30
 * Moved all the schematic and board layout out of the mech repository
   here

DONE 2019-04-29
 * Implemented register protocol

DONE 2019-04-28
 * IMU not starting reliably was a result of using read_us when the
   old code was expecting a 100us clock tick.
 * Verified laser, agitator, AEG and gimbal motors all work properly.
   Had to turn down the kd term some, probably because the IMU is
   mounted closer to the motor.
 * Verified it is running at 800Hz already

DONE 2019-04-27
 * Got I2C and SPI working
 * Got analog sampling working
 * Got BLDC outputs work
 * Got agitator and weapon outputs to work