I like your "separate" analogy. The way I always understood these things is with a black box approach. A DAC device that is meant to convert D to A, is asked to output 2V (just an example) with its RCAs and 4V (oversimplification) with its XLRs. Now, how it is achieved internally, a DAC device user should not care. Some DACs will achieve it with two separate stages, one we call "D/A conversion" and another "output stage". Others, will have a "beefier" D/A conversion implementation, that already outputs those 2V (or 4V), hence no need for a separate output stage. If pre-amplification (volume knob) functionality is implemented, these two could theoretically implement it the same way (in the D/A) or if the output stage is separate, volume control could be done there. Some designers believe the output stage is more important than the D/A (or their technical expertise lies in analoge domain, like XI Audio Sagra DAC that simply uses Soekris R2R dacs), and others will implement more of this in the D/A conversion itself, like Chord does. Do you find my understanding accurate? Want to learn, thanks!
Yes that's how I know it to be.
The only thing that I would like to add is that I only know of two manufacturer that doesn't have a separate output stage and that is MSB (select ll) and Chord (more can exist, but they are very few).