Where the research relevance comes is later, when I'm looking at the drafts with my "researcher" hat on. I can see that the first draft wasn't "right", and I can see where I recognized that...and NOW I can apply some analytical thinking to what I did. But in the moment, it's all about the creative. It just points a spot for later-researcher-me to take a closer look. Without that note, however, I don't have a clue what happened in the writing or revisions.
I think it's a lot easier for writers to do these in situ notes - it's more organic, and kind of a natural part of the process anyway. But I take a lot of notes in my photography as well - exposure times, shutter speeds, ISO settings, f-stops, time of day, etc. That's already part of an in situ log! Again, it's just the ethnography notion - providing some sort of observable record that you (as the researcher) can look at and analyze later.
Of course, I'm not at all saying that anyone HAS to work this way. I'd love to see alternative methods and protocols posted here. What works for me and for one project isn't going to work for everyone for every project. By pulling together a resource database with different methods in it, we can acknowledge all the different ways of working, and offer multiple approaches through various models.