This one kept on coming up in my searches for 'prior' art. The series connection is similar to emitter switching of a transistor with a mosfet but it's not so 'elegant' because the upper device drive has to float otherwise you kill its gate.
The parallel connection is the other one I kept on finding. Notice the time scales. In the figure you mention the transistion is being effected over
4uS..... I don't know what devices they are modelling but that is a HUGE amount of time. Elsewhere figures of 2uS seem to be the norm.As I say the problem is you are waiting for the charge to recombine so although the method 'works' you don't gain much or anything in terms of operating frequency. If you actively remove the charge as suggested by applying a reverse bias then you can reduce losses and increase operating frequency.
Cheers
DNA