by 'version 7' progs, I *REALLY* hope you refer to a fourth type of 'version 7'.
the fteqcc-v7 implies a whole load of new instructions, as well as some other twiddles.
the qfcc-v7 implies a whole load of new instructions also, as well as some other stuff that I've not read enough to explain.
the kkqwsv-v7 extends only the dstatement_t structure which means saved games are broken (fine for deathmatch-only, but fatal for hexen2).
So I really hope its a fourth version of v7.

that said hexen2 has slightly different opcodes+crc, but hey...
tbh you shouldn't really need 32bit indicies anyway. fteqcc can compile h2mp down to only 15k globals. I guess it depends how many arrays you have.
If you do try want to use fteqcc for hexen2 code, you should use the -h2 argument (which outputs h2 opcodes+crc+enables the thinktime keyword). Should work with any engine that still supports vanilla/mp progs.
Regarding dds formats, the claimed patent is on s3tc/dxt textures rather than dds itself. Its the drivers that decode them rather than your code, so its only the drivers that need the patent. So long as you don't (pointlessly?) decode it to rgba, and your gl driver actually supports it, you're fine (provide a separate tga-only version or something).
You can support RGB(A) dds textures just fine. They should be good for cubemaps etc, its only s3tc/dxt, not dds files themselves.
.