It's hard to give a complete answer without having some internal diameter (ID) measurements from the shell. You want to know whether you're dealing with ovalisation, too small of an ID, or what. If it is an ovalisation type problem ("ovalisation" may be somewhat of a misnomer because if so it was probably made that way as opposed to being an acquired problem) then you want to know whether you're dealing with some spots being too small while the rest is on spec, or some spots being gappy while the rest is on spec, or perhaps some of both.
If those measurements revealed some spots where the ID numbers were too low while the rest was good, then reaming and facing may solve it completely.
If the measurements showed the ID was too high in some spots, causing the press fit interface to not work as it should (i.e. the cup contacting the tight part of the oval and being loose or gappy in other parts), you might try an approach of building up the bore with bondo or epoxy and then reaming it down to perfection.
I suspect but don't know it's this second possibility you're running into. Loctite 641 has a gap fill spec of .1mm, and the interface wants an interference fit of another .1mm on top of that to work right and not creak. So in other words, you can approximately think of it like if there's any spot in the ID that's larger than the cup OD itself, the Loctite might not cut it.
There's debate whether using reamers on all-carbon frame parts is a good idea. The general concern, as I understand, is along the lines that the cutting operation may sever a carbon fabric layer, resulting in the frame being weakened. Some sources, Barnett's for example, say don't do it. I asked Park specifically before replying to this whether putting their #744 reamer through an all-carbon shell would be a good idea. Calvin Jones replied to me with the following:
The 744 can trim carbon fiber material without issue. Carbon fiber is not like a pair of blue jeans, where a thread is pulled and things come apart. Much like cutting a carbon fiber fork column, the resin holds things together.
I do think there's probably some kind of hypothetical risk of hurting the frame by reaming an all-carbon shell, but I think it's where I would tend to go in a case like this if the ID measurements indicated it might be helpful.
In BB30-land, special oversize-OD bearings are available that can solve press fit problems when nothing else does, in the case of bores that are larger than spec for example. I don't think anything like that exists for BB86/92.
I don't like that if the epoxy plan doesn't fix it, or only fixes it for a while, you've got a huge mess on your hands. But creaks indicate movement and epoxy is often pretty good at obliterating movement, so maybe it would work great. It's not what I would do first from what you describe.