L1P Phase2 Bugs9

1. Bug fix for focuser reset position after curvefit failure. Applies when walk=Fixed Steps or CFZ Shuffle. If curvefitting fails on the last step then autofocus is aborted and retried. Bug was sending focuser to position -1. Fix leaves the focuser at its current position and allows the Autofocus abort processing to reset it to its original position.
2. Full Frame, Sub Frame and Auto Select Star tooltips updated to be a bit clearer.
3. Bug fix for LM solver where useWeights selected and datapoint contains 1 star. In this case the "scale" is zero (which is correct) but the weighting (1/scale) or (1/scale^2) is infinite. This does not make sense and breaks the LM solver. So for zero scale, robust statistics now returns 1e-10 rather than infinity.

Testing complete.