Thank you for all of that. We have both learned a lot. Bottom line is this:
Roku
Nothing particularly wrong with what it puts out. I could quibble and say HDMI4 seems not to be HDMI2.1 (spec says only 2 HDMI2.1 ports) so should not really be using that HF-EEODB. In Compatibility mode you lose HDR10+ (for some reason) and 10-bit 4k60 (as expected). 4k24 should play with 10bits at 4:2:2 (Vero should automatically switch to 4:2:2). In Compatibility mode there are only 2 128-byte blocks in the EDID.
Vero
Vero is doing what it should except for that weirdness with the CEC addresses. I’ve found out why that is and corrected it - will be in the next update. It shouldn’t affect anything else - there are no related issues in the EDID parser.
T-500
This is making mistakes. In Standard mode, it is not passing through the Dolby audio VSDB (maybe unnecessary?) or the HDR10+ VSDB and it is not reading the DTD in the second extension block properly. In Compatibility mode, there’s no corrupt DTD because there’s only one extension block, otherwise it has the same fails as when Roku is in Standard mode. I suggest you have a moan at ToneWinner.
Final note: edid-decode is reading the EDIDs for both the Mac internal display and the HDMI attached display, then it says it doesn’t understand the first block of the HDMI EDID. You could have a moan at the author about that. If you’re interested, you could have look at him explaining why EDIDs are such a mess https://www.youtube.com/watch?v=IQlCGC_zSCE