This is an answer to the part of the question where the OP is asking if the image of the function
$$R_{2,\varphi}: \mathbf N^+ \to \mathbf R: n \mapsto \frac{\varphi(2^n - 1)}{2^n-1}$$ is dense in the interval $[0,1]$. The short answer is yes, some more details follow.

This week, I met Carlo Sanna in Turin during a meeting of number theory and mentioned the problem to him. He remembered to have read about the same question in a paper of Florian Luca. Today, I got an e-mail from Carlo with the full reference:

F. Luca, *On the sum of divisors of the Mersenne numbers*, Math. Slovaca **53**, No. 5 (2003), 457-466 (EuDML link).

More precisely, see point ii) of the theorem on the bottom of p. 458.

By the way, Luca points out on p. 459 that:

his method works for any multiplicative function $f: \mathbf N^+ \to
\mathbf R$ for which "there exist [...] $c > 0$ and $\lambda > 1$ so
that $f(p^a) = 1 + \frac{c}{p} + O(p^{-\lambda})$ holds for all prime
numbers $p$ and all positive integers $a$", on condition that $[0,1]$ is replaced with the interval $[\liminf_n f(n), \limsup_n f(n)]$ (the question in the OP corresponds to the case when $f = R_{2,\varphi}$);

the conclusion remains true if the sequence of Mersenne numbers is replaced with a Lucas sequence subjected to some technical conditions.

Among other things, Luca's proof makes use of the Siegel-Walfisz theorem.