This is a rare post where the question contains the answer I struggled with this for a few hours, but a session with GTP helped me determine the root issue and probably the cause. Interestingly enough, I also had an issue at the same time where a wire came disconnected. Lesson is, never rule out two possible issues when things don't make sense. Documenting it here as no doubt I will run into this again and forget why
Things were working fine and my supercollider scripts were failing. It would only get to:
Booting server 'localhost' on address 127.0.0.1:57110.
Found 10 LADSPA plugins
Server 'localhost' exited with exit code 0.
I tried supercollider samples in the IDE and same thing. Those didn't work either, so I tried c++. No problems, but no sound, this is when I found the wire issue. After resolving that problem, c++ worked, but SuperCollider still failed. I determined that scsynth is called automatically. I ran sclang manually. This page was helpful btw: https://github.com/supercollider/supercollider/blob/develop/README_BELA.md
This is when I saw a more specific segmentation fault error. I could have used the gdb message to get more info I think, but I didn't have symbols. Next I tried the -V 5 option. Actually, the help only talks about zero or lower, but well I assumed, and using 5 I saw a line "Loading synthdefs from default path: /root/.local/share/SuperCollider/synthdefs". Checking that dir I found some files, one call ..scsyndef and bin.scsyndef. I moved them both out and now it runs fine!
I was messing with writeDefFile, which creates the scsyndef files. According to bing GTP, it says that it can write file there as a default location. I'm guessing that's what happened as this was the most recent thing I was working on.
In case it's interesting, here is the chat session: https://sl.bing.net/c9TI0exOha0. I have no idea how long it will persist, but I learned a lot