The Mozilla bugs you link are a bit of a red herring. They refer to an abortive attempt by Mozilla/NSS to have a 'shared system database' in sql:/etc/pki/nssdb. The idea is that applications specify that as their NSS database and although it's obviously read-only, it automatically adds the user's database from ~/.pki/nssdb as a writeable token. This gets a step towards consistency for all NSS-using applications — but as those bugs note, not even Mozilla's own products are actually using it. You should support that anyway, but it isn't the focus of this bug.
The fix here (which has been working in Fedora for years, since you ask for existing approaches) is to replace NSS's built-in trust root module libnssckbi.so with a symlink to p11-kit-trust.so. Then you get the system's configured trust roots, instead of whatever's hard-coded into that particular instance of libnssckbi.so (and you're shipping multiple potentially different ones of those!)
The Mozilla bugs you link are a bit of a red herring. They refer to an abortive attempt by Mozilla/NSS to have a 'shared system database' in sql:/etc/pki/nssdb. The idea is that applications specify that as their NSS database and although it's obviously read-only, it automatically adds the user's database from ~/.pki/nssdb as a writeable token. This gets a step towards consistency for all NSS-using applications — but as those bugs note, not even Mozilla's own products are actually using it. You should support that anyway, but it isn't the focus of this bug.
The fix here (which has been working in Fedora for years, since you ask for existing approaches) is to replace NSS's built-in trust root module libnssckbi.so with a symlink to p11-kit-trust.so. Then you get the system's configured trust roots, instead of whatever's hard-coded into that particular instance of libnssckbi.so (and you're shipping multiple potentially different ones of those!)