I would like to apply it to mainline, but it will be long process with some probability of success
It's pretty simple - 2 patches. First should update device tree documentation with new compatible (A64 specific one followed by A13 one) and your patch, updated with double compatible. Both patches should of course have proper title and body and accompanied with cover letter. Third patch can be added with "DO NOT MERGE" tag, where you add IR to pine64, to show how you tested that it works.
This could be great for first time Linux contributor, if you want to go through process yourself. If you don't want, then I can do it instead. Anyway, I don't want carrying patches for too long, especially if they can be easily upstreamed.
Actually Pine board has soldered IR header, where you can just insert IR diode. I did it some time ago and IR start to work with original BSP without any further manipulations. At the moment I can't see any drawbacks if we leave IR enabled, but IR diode will not be inserted.
True, but you can say that for other peripherals too (like pine64 wifi) and suddenly there can be a lot of patches which can't be merged into Linux. Linux has a policy, which can be worded like: "If it is not soldered, then it shouldn't be enabled and empty headers don't count.". If you would submit such patch to Linux, it would be rejected with explanation it should be done with DT overlays.
Because I don't want carrying patches forever, I have to go with Linux policy. DT overlays should be already working, but I'll try to find an easy way for handling them. Ideally, it should be as easy as Raspberry Pi DT overlays.
It's not supported by hardware. I'll change it to be decoded in software but it will be probably too slow to be useful.