You can tweek PID constants forever, chasing changing payloads, weather conditions, & speeds. At least, they converged on values that would wear the tires out acceptably & go reasonably straight. It's not going to be perfect. The highpass filtered angular rate was worthless. It's amazing how for you can go with just highpass filtered heading. The next step would be adjustable bandwidth.
Note how the FIR filter isn't an exact phase shift of the error. Just fishy how it needs to be 7x the P gain.