I figured I should try this for completeness!
First, two of the images from the above image set with the preset applied to the rgb ratio versions compared with the preserve hue = 100% version.
That is indeed better but there is of course still a difference. It’s really hard to see how they differ exactly on a photograph like that. I personally really love synthetic tests on these occasions where a simple sweep can show a lot of what happens! Again, try to imagine that this generalizes to any curve shape and norm of filmic, there will mostly be a change in when and how fast it approaches white. The default contrast setting of 1.6 matches fairly well with the color balance rgb preset. The input fills the rec709 gamut and the working space is rec2020.
I hope it is clear to everybody that these two methods do not produce the same results. I do not even want to call them similar when comparing the synthetic tests! I would like to be so bold as to say that it’s impossible to reproduce the look of per-channel mapping (with preservation hue on or off) with a combination of an rgb ratio map and color balance rgb. Scroll up to post 477 and think about how you would warp the image data before an rgb ratio mapping operation such that the combined effect is the same as per-channel mapping.
Simply put, do you want the pre-channel look? Then use the per-channel method.
One reason for this is that the chroma changes that happen when applying a per-channel mapping are “hyperbolic” and never clips the gamut which is a very nice property. I think this hyperbolicness would be extremely interesting to use for a color editing module but that is a story for a later date.