Alternatively, until homebrew is updated you can run brew edit emacs and change the url and sha256 to match 24.4, then run brew install emacs with whatever options you would like (brew options emacs for a list). I haven't tested this yet, but it should work, and I will report back after I've tried.
edit: So this did work, I used the .tar.xz with sha256 47e391170db4ca0a3c724530c7050655f6d573a711956b4cd84693c194a9d4fd. One thing to note is that the old 24.3 stable formula has a bunch of patches applied that are no longer required, so you need to edit those out as well (this area: https://github.com/Homebrew/homebrew/blob/master/Library/For...).
For 2 or 3 years now, whenever I have compared the two, the current version of "Mitsuharu" Emacs has always had fewer Cocoa-related bugs than the current (release) version of "FSF Emacs" (i.e., Emacs without the professor's patches applied). I am writing this comment because this Mitsuharu fellow seems very reticent about promoting his own (excellent) work.
True, I didn't find any notes on what key to use and the key I typed in is the one the signature itself said was missing (so it is in fact a bad choice). Your criticism is valid. Sorry if I misled anyone, I meant to fix that but I didn't find much guidance on the gnu site as to what signatures to load to bootstrap trust.
Just a note: I think I was hoping (but didn't confirm) that I was downloading the key from GNU and the file from a mirror. So I thought I had the minor protection that I was at least safe from somebody who could only alter the mirror. But I agree with the criticism: not being clear what you are claiming to actually check in a crypto-situation can confuse others and cause harm. Sorry about that.
No worries, end to end PGP leaves a lot to be desired, but I feel it's important to point out when the common recipe steps fail to protect against exactly the adversary one would think they do.
1) Download from http://www.gnu.org/software/emacs/
2) Confirm: 3) Build (from INSTALL): 4) run: