I am the latest member of YANG Doctors to review this document and the modules therein.  I looked over chopps' previous review, and it appears most of his comments have been addressed.  In my own reading, I found inconsistent use of capitalization and punctuation in descriptions (e.g., some ended in periods, some did not; most started with a capital letter, some did not); as well as inconsistent quoting.  All modules would benefit from a `pyang -f yang` normalization and an editorial pass.

In the ietf-lisp module itself, there are a couple of patterns in there where I wonder if the regex is what you want exactly.  For example, is it okay if an eid-id starts with a ':' or a '-'?  For the locator-id, this is a string of 1-64 characters, but the regex hints it could be zero or more of the character class (a similar example exists with hop-id in address-types).

All modules' initial revisions reference the original LISP RFC but do so with a URL only vs. a more correct, RFC 6830: ... syntax.  And speaking of revision, most of the modules have a revision of 2021-02-22 with the exception of itr and mapresolver.  This isn't a big deal now, as I assume you'll set all of these when the draft is finalized.  You should also update all of the copyright years and copyright text.

As to the two questions asked here, I can see some benefit of breaking out the IANA parts of address-types into a module that they maintain.  But in its current form, I don't know that it makes sense to have them maintain it.  As for geoloc, I do see some overlap, but I am not a LISP expert at all, so I cannot comment as to whether bringing that whole module in makes sense or would even work with LISP implementations.  That is, it seems LISP lat and long are expressed in degrees° minutes'seconds" whereas geoloc does this as a decimal64 from a reference frame.  I do feel that whatever direction is taken, text explaining why geoloc is not used is useful.