That's just phenomenal, congratulations! So happy to hear about your excellent results!
Nice "surfing," well done! What you've described above is what some closed-loop systems are attempting to do automatically - they use temp basal rates not only to adjust for activities, sickness, or such, but to make meal-bolus corrections based on CGM data. In a manual system, such as the current OmniPod, using temp basals for meals is just fine, except for one small problem, which is related to your question about IOB. IOB calculations on the PDM take only boluses into account, and are completely ignoring basal and temp basal rates. If you wanted to stick to the conventional bolus/basal model the PDM assumes, you could use extended boluses instead of temp basals. Lets say you give a 1U meal bolus together with a 50% temp basal increase from 0.2U/h to 0.3U/h for 2 hours. This temp basal amounts to (0.3U/h-0.2U/h)*2h = 0.2U of extra insulin for that meal. In this example, you could reproduce exactly the same effect by issuing a 0.2U bolus extended over 2 hours. Then, just as you may abort a temp basal based on CGM reading, you may terminate an extended bolus at will. An advantage compared to what you are doing is that the PDM would take the extended bolus into account in its IOB calculations.