M2 U M U2 M' U M' U2 M' will solve that case, fyi. (This is just MU U perm (M2 U M U2 M' U M2) cancelling into insert (M U2 M').)
General consensus is that this is not worth it; if you look at how the MU U perm algs work, or the common MU H and Z perm, algs, they're all essentially doing 4b then 4c. You're not skipping 4b/4c by setting up to EPLL; you're just doing it less efficiently.
Thank you for your response! With the H and Z perms I noticed that they were essentially doing 4b / 4c. I did not know the MU U perms (only learned the 2-gen ones when I started with CFOP). But playing around with this helped me notice that I have an inefficiency in my current way of solving some 4b cases. For some reason, I would solve a "scramble" such as U M' U2 M (UL Edge between opposite colours) not simply with M' U2 M, but with the inefficient M U2 M' U2 M2. This might have contributed of my intuition of the U perms being faster.
Thank you!