All you've really done with all of the above points is explain the concept of risk-averse indices, on which I have a whole section, in chapter 13. None of your questions can be answered unless or until you enunciate a bet spread and ramp for a particular game (whose rules need to be specified in great detail).
You can't use the BJA3 charts you allude to for a counter who spreads, because all the doubles you discuss become much more valuable as the count increases and the bet increases proportionally. So, if you don't use r-a indices, and you make all the doubles at the e.v.-maximizing indices, instead, your SCORE will, indeed, be lower than if you were to use r-a indices. But no, absolutely not, your SCORE will not be higher if you don't soft double at all, instead of using r-a indices. That just wouldn't make any sense.
Same type of answer for the insurance question. Insuring naturals lowers variance -- that's a given. But, it also loses money at Hi-Lo TC of +1, so again, you would have to simulate your precise betting ramp to see what SCORE would be by insuring more frequently (and, therefore, incorrectly). The reason studies like these aren't common is because they produce different answers for each of the conditions that prevail for every one of the charts in chapter 10 ... and more!
Don