These flags will later be used to test which calibration has been done already.

We can use the same flag to optionally set settings, rather than use NAN to indicate nothing set.

Since this is done in quite a few functions, this should make the build size smaller and the code easier to follow.

Since this is done in quite a few functions, this should make the build size smaller and the code easier to follow.

When the hub first boots, it is good to have an initial value for the bias to help suppress drift during the first nonstationary seconds.

As an estimate, we can use the bias first saved on the previous boot.

When starting from the latest sample as we did before, the odds are higher that we start on an outlier and have to restart, or allow larger errors.

Now that it is a bit less sensitive to outliers, we can reduce this value. See pybricks/support#1105