This change only affects the linear focus algorithm, and rev's it to 3.2.
Previously I've found that the algorithm finds solutions, but stops a little too early.
Now, once it finds a solution within tolerance, it will continue to focus-in and see if a better solution is available
if (a) the HFR has been decreasing, and (b) if the next step is less than the minimum position of the polynomial solution
found in the first pass, and (c) it is not near constraints like maximum number of steps or minimum focus position.

Basically it's running the linear focus algorithm with standard parameters (e.g. in the simulator with step size about 1000
starting from around 40000) and noticing that in the 2nd pass, depending on the samples, it does not take the first acceptable
solution if it's been decending and is not yet near the polynomial minimum.