Agree broadly with your proposal, RC
But would have a couple of variations :
Single-pilot IFR in a light piston, particularly without A/P, is the most demanding variation of the genre - 35 hours is not enough training. If anything it should be more than the CPL equivalent since most new-minted CPLs will be closely supervised and "Captained" for a very long time.
A "test" every 13 months? I prefer the FAA "review" - do it till you get it right rather than one strike and you're out.
Allowing training in the subject's own machine would be the single biggest cost-saving mechanism - the approved school lobby would probably squash that one.