I never used MAC Lisp, but I read the documentation for it. CL is more similar to it than ELisp. It also didn't come particularly later. It's common to associate the date of ANSI CL standard publication with the "birth" of the language, however, the standard in this case was the attempt to unify and codify the existing language that was implemented in slightly different ways by various groups. You could even think about it as incorporating MAC Lisp as one of the components that went into standardization. Eg. the loop macro in CL is almost exact copy, whereas ELisp only has it in cl package. There are few more things like that.

If memory serves, the idea for ELisp was to eventually become the "proper" Lisp for system programming. At the time, people used the word "Lisp" very liberally, in a way how we today might use "high-level programming language". And, in a way, Guile was supposed to become that, but as is the case with some such initiatives (eg. Hurd) it never gathered enough momentum to replace the predecessor that was "good enough" and was continuously kept afloat with more and more upgrades.

As for the larger set of complaints about ELisp... it was good for a very long time, but it's getting old. There are many programming practices that are difficult to incorporate into the language. There are what we collectively came to believe to be programming anti-patterns that are encoded into the core of the language. And it's hard to do a proper face-lift because it will inevitably break a lot of old code. Also, doing such a face-lift you'd always be perplexed by the idea of what if a few years from now you could do an even better job of rewriting it into something even better? Knowing that one such change will put a ring on you, perhaps for the rest of your life, it's hard to make the step. Also, there are and have been many attempts of moving away from ELisp, and none really gained much traction among Emacs users. So, it's hard to imagine that the next such attempt will succeed.

My fear in this regard is that one day, instead of evolving and replacing with something better, we'll be reduced to using junk like VSCode due to some underlying system aging out completely. Similar to what happened to Firefox when it lost its ability to house extensions. I pray this day never comes, but, in practical terms, it means that at some point the community of ELisp users needs to cut losses and embrace the loss of older libraries to move to a revamped ELisp2 or w/e it will be called.