Unlike insulation dynamic mass is difficult to quantify since it works under a different mechanic with lots of variables vs. steady state U-value or R-value that you find a lot of info on like min thickness per climate. Dynamic mass should be ran through a dynamic processing model like WUFI with better chances to get it right, and then verified by test. At a minimum I'd suggest calculating a u-value (thermal conduction) for your climates max cold temp differences, the inverse of that is r-value, then I reduce it by 30% for mass benefit to get a min thickness at the core insulation. Guessing is risky. You do not add soil stabilizers unless your soil type needs it, nor rebar unless in a seismic or wind zone that requires it, and I'd never use steel. The other variables that determines thickness is compression, flex modulus, and deflection. Thicker the better for getting wide load distributions to the ground especially if on weak soil with low compression and shear.
In the US goggle NM Rammed earth code or call their building and safety office for the latest and/or amendments. Follow it, it has some good guidelines but be careful it is more for SW climates. 18" is minimum, most use 24" with a R-10 insulation core as a thermal break. I would not use foam, I would use Roxuls IS Board @ 3" (staggered 1.5") for R-12 core. I have success using a siloxane (sand/natural silicone sealer in wet climates). All these thicknesses include insulation or what is referred to as "whole wall r-value and/or DBMS (Dynamic Benefit Mass Systems" " that includes the insulation, thermal mass, and any other layers. I'd run a test in the climate and check for thermal bridging. There are many cases of sucesfull CAN rammed earth like SIRE you could contact for help:
http://sirewall.com/ some with little to no HVAC load in freezing temps, or net zero and off grid.
I'd also suggest reading
books on the subject there is a lot to know.
Hope that helps, good luck!