I don't think it likely that expert engineers get overruled by marketing on such straight forward issues.
let me give you a historical example. around ww2/cold war stalin reverse engineered the american b29 bomber into their Tu-4, they even copied a repair patch on the plane. the engineers MUST have known that from the material they were copying, that that was a repair patch , however they were so afraid of stalin, that they copied it exactly, just so no one would say, hey that feature wasn't there.
engineers get overruled all the time by marketing/ leadership, not to produce the best engineer, but to make the most money or some other equality vapid goal. this is extremely common.
With respect, I'm not sure you've understood marketing, and its role in product design. Kawai marketing are unlikely to carry the fear factor ascribed rightly to Stalin.
For one thing, market research precedes R&D. First, customer tests are conducted to determine what the new instrument should deliver compared to the old (in this case, how to improve the RHII action for the new ES8). Second, the engineers are tasked to draw up plans to meet the design brief. They are only going to be Challenged on a final design if it fails or exceeds the brief.
Why do you think that the marketing people would wish to overrule the engineers choice of action? It makes no sense to me that they would, as each new action already follows the marketing brief.
The engineers would be briefed to create let-off simulation for the ES8's new action (as customer research suggested it). The engineers would then draft design, test, bug fix, test again, then get sign off that it meets standard. Likely, they modified the existing mechanism: the RHII.
If later, the majority of post-market research said: we don't like the let-off, please get rid of it, then the engineers would get a new brief for the ES920, with the note: remove let-off simulation.
As that didn't happen, we can conclude that the post customer Market Research showed that the action was generally well received, with the majority if users not complaining about the let-off simulation. Don't forget, the ES8 was the most popular portable of its time, winning awards etc.
From the post market research on the ES8, likely there were slight technical issues identified that engineers were asked to address ie, to make the RHIII more robust for the ES920. Essentially, the RHIII action was fitting to the ES920 brief ie, as being a piano-like action that can be used in stage and portable piano. It would thus need to be cheaper to produce than the GF 1 action (as it'sa lower model) and a bit faster and lighter too---so stage pianists would be able to play organ and synth licks).
Thinking about the brief for the ES520: it would likely have been to generate a model midway between the ES110 and ES920: with an action not as premium as the RHIII---hence lack of let-off simulation---but better than the RHC action in the ES110, hence the RHCII!
The fact that some of it's characteristics eg, speed of key return etc., might be to you more useful to your needs than the chaaracter of the RHIII action is irrelevant to Kawai, as you are one customer in a larger group. The ES520 RHCII action will not have been designed to outperform the RHIII in terms of how closely it simulates an acoustic. That is the point I am making. That you prefer how the RHCII plays is idiosyncratic to your perspective.
So the engineers don't have one opinion and then get overriden due to the fearful marketing department head; rather, they follow a brief and design to that. Only if engineers and R&D people totally leave the brief would there be any feedback to change something.