Just using the journal template and looking only for compilation errors does not produce particularly good results. If I am preparing something for a journal like Quantum, where I am in charge of typesetting my work, it would take me a day or so to ensure figures are well placed and margins are not weird (in other words, not simply make it compile, but make it compile without warnings (not errors)). Most scientists do not care about the warnings latex emits as long as it compiles, and the results are ugly papers. I find it reasonable to pay $1000 to someone to take care of that processing (if they do a good job at it, which APS's and Nature's journals do).

If you go for high quality and work on the figures is performed, my (unreliable) guess is half a day for a professional designer. It takes me much more, because my latex skills are limited and because I insist on using the most standard and reproducible tools/packages in latex.