Possibly, the partial success in the question is already the most success one could get. At least I found no further proof that this error indicates missing logs:
I/neqo_common::* [neqo_common::log] Logging initialization error SetLoggerError(())
Maybe it is a bug, or maybe that message was just carefully crafted to confuse unsuspecting would-be log readers. Either way, this would leave the following configuration entirely sufficient to get all the messages that could be emitted from the (Rust) components handling those transports:
about:logging?modules=timestamp,pipnss:5,cert_storage::*:5,nsSocketTransport:5,nsHostResolver:5,nsHttp:5,neqo_http3::*:5,neqo_transport::*:5,pkix:5
Using this URL is even better than the environment variables, because it leaves out all the unwanted mozilla.com connections from the beginning of the log file.
I am starting to suspect the logs I am looking for would be there and be marked SSLGetClientAuthDataHook if only the relevant rust component was to set that up. I may not be seeing those log lines because the thing I want logs from is simply not called for that protocol...
One alternate way to generate logs is is opening up https://profiler.firefox.com to use that sites special permissions to add the profiler button, press that button, load the page a few times, press the button again. That will open a new tab which can show samples matching commma-separated keywords in function names when used via Parent Process > Socket Thread > Call Tree > * > Filter stacks: neqo,nss - Unfortunately, even with the minimal 0.01ms sampling interval, on a reasonably fast machine this will only snapshot 0-1 relevant function calls per request, producing a rather discontiguous & incomplete picture.
