you demonstrated that there is a lot of slop available in processing photographic material.
And that depends on how stringently you're examining your results.
I have never done sensitometry, but I understand how it works (and I'm gathering the tools to start). I'm pretty confident that someone who routinely exposed and processes test strips, and then measures the results against an established (by previous testing) curve will be able to spot a 3% variation in either developer strength or component mixture. In many cases (say, with Xtol, but more generally with most phenidone-based developers) a component of crucial importance represents much less than 3% of the total solids to be dissolved. Use a developer that comes in two or three bags, and you add another variable.
If consistency is the bane of your existence, you may as well "pinch and throw" starting from individual chemicals. I'd bet I could make Caffenol that way and get images -- but if I want all my films to be developed the same, so they print the same and I can tell if I'm over- or (less likely) under-rating film speed, I need to avoid a condition that could halve the quantity of one of the developing agents, significantly change the pH of the solution, or otherwise make a change that in fact makes a difference (or that I
don't know won't make a difference).