sorry if I have missed something here, who knows.
the format spec has certain "omissions" which are more than a little annoying for those of us doing custom implementations.
so, a few noted issues/requests:
"UTF-8" scheme could be defined in the spec;
make more obvious what is and what is not byte aligned;
actually explain the rice coding scheme (in particular, that the unary-prefix is 0 based and 1 terminated, vs the inverse, and also the sign-folding convention);
note that the left/side, right/side, and mid/side encodings, may involve additional sample bits (this one was particularly frustrating);
make it a spec that one can look at, and get a working implementation from, and not have to guess things, follow (broken) links, or pick apart existing implementations to figure out (or spend hours or days using trial-and-error to beat through some parts of the format, all the while, not having any real idea exactly in what part of the format things are going wrong).