The 9000 series algorithms are protected, so we can’t peek inside how they did things, and they may use tools outside of the VSIG toolbox.  One thing to try might be to use one of the pitch shifts ahead of the sampling delay (such as diatonicshift, Pitchshift, reverse or stereoshift), say up 1 octave and then have your baseline dlysmp2 at 50%, to get the original pitch back.  This would let you double the pitch, but there would be some dissociation of pitch and speed and likely some loss of fidelity.

Another option might be the bucket brigade delay, which can impart pitch shift, but you will certainly lose fidelity (by design).  There might be a way to do this in RNBO, but audiobuffer isn’t supported for the H9000 yet, so you’d need to do it in the ram based delays.