Hey All!
Thank you for your response @morph027.
Well after many hours of trouble shooting, one setting at a time, I’ve solved the issue. And it was a simple and stupid one on top of it.
And since someone else might have issues getting this going, here is what I did to narrow it down.
Troubleshooting
Step 1 - Upgrade Software
Thinking along the same line as @morph027, that it might be the version of Janus/Signaling backend that might be causing the issue.
I began trying to update both Janus and the Signaling backend to the newest version on the Debian LXC Signaling Server container. Which proved to be more difficult than on might think since there was no repositories available and manually installing Ubuntu packages didn’t go very well either do to many missing dependencies in Debian.
Not wanting to build packages myself, I decided to spin up a Ubuntu 20 (Focal) container and installed all the software again using @morph027 repositories, and copied all my configs from the original container.
This did not solve my problem, still got the same error, even though all the software was at the newest version.
Step 2 - Docker
Being really puzzled and annoyed at this point, I decided to install a docker release of Nextcloud AIO in my home lab. To test the Talk/HPB setup, but mostly to look at all the config files and see if there was any difference. The docker instance worked as expected, without errors, in my home lab.
So I copied all the configs from the docker Signaling instance to my new Ubuntu LXC Signaling container. AND… The same problem continued.
Step 3 - Which Server is the issue
Being even more puzzled and annoyed at this point, I decided to figure out which server was the issue. Was it the NC Core LXC container or the NC HPB LXC container.
The quickest way to do this was to point the my new docker instance to the Ubuntu HPB LXC container. Since I copied all the configs it was just a matter of changing the domain in the Nextcloud docker instance and changing the backend domain in container.
To my surprise, it worked. No more error. This confirmed that, Ubuntu LXC HPB container was working properly.
Step 4 - What is wrong with my NC Core LXC container
Being even more puzzled but less annoyed since I was making progress, I stated looking at all the differences between the docker NC Core and LXC NC Core.
At first I thought many the Talk app, did not upgrade properly from a previous version or had some old files that where causing the issues since the LXC NC Core had been running for a few years. I manually removed all the files and settings using the CLI, then reinstalled and setup the talk app. To my disappointment this did not work, same error.
Then thinking that there was something wrong with my NC Core installation, I spun up a new LXC container and manually installed Nextcloud and pointed the Talk app at the existing Ubuntu HPB LXC container and added an extra backend for my new LXC NC instance to the HPB. To my amazement, everything worked. No more error.
Conclusion
Still puzzled, why the new instance works and the old one does not, I began looking at all the differences and ruling them out one by one.
Was it IPv6? Since my new NC Core instance did not have IPv6 assigned yet, I assigned an address and tested… Yup everything still works as it should.
At this point it was down to a difference in NC Core configs or a bad upgrade on the original container. So I started looking at the configs, and started coping all the configs from the old container to the new container and testing as I added configs for Memcahed, Redis, Email, Etc. To my surprise the new container continued to work properly… No errors… But the old instance still did not work.
At this point I only had one section left in the configs, trusted domains. In my old instance I had 3 domains (a FQDN server name domain, and two user friendly domains) and in my new one I had one (FQDN server name domain).
Light bulb moment… Was my old NC Core instance using a different domain when contacting the HPB instance? So I added every domain, in my old NC Core instance as a separate backed in the HPB instance. And to my amazement and joy the talk app on the old installation worked! (Had a drink at this point to celebrate. 
)
So the moral of this story my friends is that NC uses the domain that you are accessing the GUI through, in its signaling to the HPB. So if you have multiple domains setup in the NC Core instance, that you are accessing the GUI through, you also need to setup those domains in your HPB backend. (I did not see this in any documentation, so it was news to me) Hopefully this helps someone in the future.