The argument is completely illogical for the point you are trying to convey. I already explained why.
Savescumming is not restricted to games with RNG. Not only that, but it is also player side issue, not a game issue.
Portraying it as a factor against the type of RNG being discussed is dishonest, to be light about it.
Note that this is not a critic to what you are pointing out about savescumming creating bias, that much is correct, but it doesn't support what you're trying to say it does, at all.
The core point is there is practically forced tradeoff between save frequency and statistical fidelity in a game with RNG. High save frequency is useful for players who are not that good at the game but also dislike experiencing the same contents repeatly, it's not ideal if it's hard to achieve without creating bias as a byproduct. So it's a design issue.
Save scumming is certainly possible in a deterministic game if it is information incomplete or reaction focused, but that's mostly irrelevant in this context.
