Doing a high quality analog print and scanning this is certainly a viable option. however, it will be noticeably softer than doing a direct scan at really high resolutions (Actually even the analog print will already be less detailed, I did many tests on that).
This downside has also an upside, as the grain texture will be less pronounced but still looking very organic.
Long story short, I think that you're over-analysing the wrong areas that impact on the psycho-optical perceptions of the image. The main reason for making a very high quality repro of a master print is to try and replicate the complex interplay of the emulsions (and, to a point the optical systems of enlargement etc) that no scan can achieve because of the physical limitations of digital sensors. Over-unity low frequency response of film and paper emulsions and faster MTF fall-off at higher frequencies in the paper emulsions seem to result in a fully analogue print that can look both crisper (higher sharpness of big objects) and yet cleaner than even the best digital scan (and that includes vapourware touted as 'superior' to anything ever commercially available) - which while it might seem 'sharper', does so at the expense of increasingly unnatural (compared to a darkroom print) over-resolution of granularity in the highest frequencies. These rather complex trade-offs seem to have possibly been somewhat understood within parts of the now largely extinct world of high-end CCD/ PMT scanner manufacturing - as you go above 6-8000ppi, you are getting into (potentially very useful if you are designing emulsions, developers etc) analytical territory, not really imaging (which is unquestionably a problem if you want to make very large prints from scans) - and the problem of digital scanning not having the same MTF characteristics as film/ paper seems to start to become apparent - where optical enlarging still seems to keep on giving good transmission of film characteristics (note, not necessarily more resolution of original object) until you get to the optimisation limits. But that has to be set against the fact that the need for the sort of roles the specialist Rodagon-G's etc were intended for has been filled in the ephemeral advertising market by other approaches, none of which really need fed by more than 6-8000px on the long side. So, it's a set of compromises, most of which derive from the assumed limits of image size needs relative to quality from 40-50 years ago.
What I have seen is that in a properly (if accidentally) done double blind test, people tend to think a big fully optical print (done with a Rodagon-G) was an ultra-high-quality scan and the high end scan was a darkroom print, because of the way they complied with the errant preconceptions people have been persistently fed as dogma.
As a side note, there's also the problem of traditional interneg/ interpos materials having to be carefully chosen so as not to introduce too much MTF boost between camera neg and print, or it would start to rapidly look unpleasant (and, yes you can make bad HDR in the darkroom).
Overall, this rather distracts from the OP's original question of whether good 18x enlargements from 6x6 are possible. The answer is an unequivocal yes, with the only note of any importance being that the perceptual granularity of the Gold 200 negs will be closer to Portra 800.
