The world was way richer back then and could afford to spend time and money building desktop scale CPUs for game consoles that sells just tens of millions over few years. They just constantly made new CPUs.
To me, the executive level idea was to spare a core for system and minor peripheral tasks - almost all game console before that generation had single CPU architectures with no resident operating system. Transitioning away from a manually coordinated, single threaded code, to an asynchronous multithreaded programming model, was a challenge in itself, and having to deal with it while operating system forced by the console manufacturer taking away control might have been too much for developers.
(Sega Saturn had 2x SH-2, Dreamcast was to have WinCE on ROM but cancelled. Sega being Sega)
To me, the executive level idea was to spare a core for system and minor peripheral tasks - almost all game console before that generation had single CPU architectures with no resident operating system. Transitioning away from a manually coordinated, single threaded code, to an asynchronous multithreaded programming model, was a challenge in itself, and having to deal with it while operating system forced by the console manufacturer taking away control might have been too much for developers.
(Sega Saturn had 2x SH-2, Dreamcast was to have WinCE on ROM but cancelled. Sega being Sega)