Allow the user to set epsilon and maxIterations for DSST mean parameters conversion
Recently, we have performed lots and lots of osculating->mean parameters conversions. And we have stumbled upon a few cases where the hardcoded values for the following parameters have proved troublesome :
Our tests have shown that the algorithm converges just fine if we increase those values a bit. Actually, increasing the epsilon is enough in our experience, but we believe it would be nice to have control over the maxIterations too, just in case.
I think it would be a good idea to allow the user to set those values, just to ensure that all cases can potentially be handled, even when convergence is slow.
In addition, it could be nice to increase the default values a bit, to ensure the algorithm converges in nearly all cases without fiddling with the values. Based on our (limited) experience, we would leave max iterations at 200, but multiply the epsilon by 100 (2 extra orders of magnitude).
If there is consensus on this issue, we can contribute the change. It should be quite simple : two extra "setters", and maybe a change on the default value for epsilon.