Dynamic Sfusion Attempt 2

[gorbit99]

(Attempt 2)

This PR makes the following changes:

IMU support

Implemented hi-res mode for icm42 (larely untested)
Added VQF params tuneable by sensor.
VQF's MBE (motion bias estimation) feature also got enabled.

Sfusion calibration changes

Moved the original "classical" implementation into separate class called SfusionCalibration (might be a confusing name, suggestions are open).
Alongside that this PR implements "nonblocking" or "automatic" calibration, where the tracker tries to calibrate itself, whenever it detects a rest period. These new calibraions are then applied the next session.
Nonblocking calibration was made the default.

[kounocom]

• Confirm that any combination of 2 sfusion IMUs works on overclocked ESP8266
• Confirm that any combination of 2 sfusion IMUs works on ESP32-C3
• Fix ICM-42688-P high-res mode
• Use the relationship between both gyro calibration points instead of hardcoded TemperatureZROChange for the TemperatureGradientCalculator whenever present
• For basic accelerometer calibration, all 3 axes can be calibrated at once
• For advanced accelerometer calibration, save the readings from all 6 sides, then once they're all present, use magneto to compute and apply the values on next boot
• Use separate rest detection instances with separate parameters for rest bias estimation and sensor calibration steps

I'll edit this to tick off completed tasks, or to add more tasks when I remember them

[Siraxa]

tried this with BMI270's and aux wasn't calibrating I changed back to
.biasSigmaInit = 0.5f,
.biasClip = 2.0f,
.restThGyr = 2.0f,
.restThAcc = 0.5f,
and it works fine, I guess Gyroscope th was a bit too small

[Eirenliel]

Can we please rename it to DynamicCalibration instead of NonBlockingCalibration? Becaue the latter is not descriptive, especially in the config enum.