I didn't filed a but with libwebrtc, since the particular changes in my patch only effect the implementation in firefox.
The patch only reverses changes which looked at the time the pipewire patches from fedora landed usefull regarding dmabuf modifier negotiation.
The problem was that if a client didn't announce a modifier and another one did, pipewire would treat the client without modifier as if it could support all. This results in in wrong negotiating via pipewire, since firefox, or in general libwebrtc only supports linear buffers (modifier 0). Thats why i opened my first MR.
Since pipewire 0.3.29 the situation has changed. There is now a flag to indicate, that an announced format option (enumformat) should only be matched to another one, if a property is set (in short: only match enumformats containing modifiers with others with modifiers and analog for those without.). This would require firefox or libwebrtc to announce two enumformats, one for shm buffers and one for dmabufs.
This "new" approach will be discussed next week at guadec to make sure all portals will handle negotiation wrt. dmabufs the same (disclamer: I'm the one adding the pipewire flag and implementing it to the wlroots portal).
This has the problem, that with the new propasal current firefox will only announce it's capability for dmabufs and not for shm buffers. The problem is here, while dmabufs can be more performant, this isn't really the case when the buffer gets mmaped. Additionally shm buffers will work in all cases and in most cases are more stable and more performant in reality (mmapping dmabufs on an external amd gpu is really slow and dmabufs are not supported on nvidia). This problem didn't occure, since all portal implementations ignored the negotiated modifier information for now.
My first patch was to implement the correct behaviour for the new proposal in firefox, which shouldn't interfere with the old approach, but that's probably better done in libwebrtc as soon as all portals give their ok.
The second patch only removes the problematic part of my first patch (removing the modifier and using CHOICE_FLAGS_Int instead of Int, which is internaly treated the same, but the choice_flags is the correct one).
I hope this answers most of your questions. Feel free to ask again, if i left something unclear.