Preload CSS from a few months ago is not evaluated by test servers, incl. and from PSI. The effect of preload loading CSS to improve loading speed is very small and not worth using.
What removes CSS render-blocking is only critical CSS. But creating a working well-critical CSS takes a lot of time and this increases the budget maintenance of the site.
For several months I have been working on a project where I move a large volume of external CSS to the end of a global template before loading a java script. I also use and create my own critical css in the head, tailored to my content and the most important css rules.
Of course, all these changes that I make are related to many tests on the site, both with different test servers and direct tests with different browsers. However, a good site visitor experience is more important than a site loading speed.
For now, the tests on my site have an average mobile loading speed of 75 and 97 for the desktop in PSI tests.
But I still have a lot of work and analysis ...
Specific advice for your site, for improvement, I can write, but if you share a link to your community.