Seems like a moderate, reasonable, scientific Approach IMO.
You’ve conducted the same test on two bikes, using the same rider and treating time taken as your measure, you’ve calculated the differences in mass, I’d say Aero is negligible for a climb (discuss) and you say you’ve maintained a similar HR for both (as good a measure of the work done as possible)…
I’d buy that 22% figure, applied to climbing, accepting an appropriate margin for error…
I supposed you could try to verify it by conducting a similar set of tests rolling along on a flat level surface, and coasting down, from a standing start. Both would probably require calculating the difference in frontal area to reconcile the aero differences…