I've implemented curved dynamic intensity into the pseudo-deviance algorithm of the raytraced dynamic light sampling. It's almost as smooth as full deviance, while preserving isolated single-point shadows. When comparing all algorithms, the curved dynamic deviance is clearly the best one, so there's no need for the others.
Also, I've fixed the issue with the pseudo-deviance at the edges of the surfaces. Together with the above changes, the pseudo-deviance looks visually complete now.
The only thing missing is optimization. I'll work on it as soon as possible, and the next version will be released when the pseudo-deviance optimizations are complete.
Meanwhile, I've been preparing a huge amount of documentation to counter the claims made against me in the lawsuit. There's still a lot to go, but it seems that I'll have some more time to produce it than I was expecting.