You heard of the 80/20 rule? This very much applies with ML raw, except it's more like the 95/5 rule.
The main advantage of shooting in raw is twofold. Firstly your image contains much more detail coming in from the sensor, resulting in a sharper image that resembles that of much higher end cameras. The second advantage is that you are given more latitude in post to do aggressive grading.
These two major improvements are gained the moment you begin shooting with ML raw (ideally on a camera like the 5D3). Any additional pixel peeping is IMHO a bit of a waste of time that even some of the most ardent professionals in our industry wouldn't be bothered with. ACR's debayering algorithm has a slight edge over Resolve's (although in about two weeks, that might be moot). The image improvement advantage that you gain is not worth (once again IMHO) the extra expenditure of time, hard drive space, and complexity to the workflow.
I generally shy away from saying things like "the general public will never notice any difference", but in this particular case I do think it's quite true. We tend to sometimes become overly fixated with these things that it becomes an unhealthy obsession. As a builder of several Hackintoshes, I know I get obsessed with squeezing every bit of performance out of the CPU as possible, but at some point I realized that it's much better to have a slower but robust and reliable machine than a slightly faster but quirky computer. I think the same philosophy definitely applies here.
