Rewrites may not happen that often, but plenty of large-scale corporate tech projects flail for years without ever reaching a functional state.

If the choice is between a sub-optimally performant but working system, or a boondoggle that never gets off the ground, the “better engineering choice” is probably the former.