- We now use the same values and formulas as in Smith & Aldrian for the detector 2D variance.
- This changes the ranges of values a bit, so set an equivalent reasonable regularisation default value