The relative risks in that study look very compatible with reporting bias. Certainly nothing in the way of a major causal input. Almost all psychological research is as wooly as this and I think tells us very little. If anything it is quite a nice negative result indicating that premorbid psychological state probably isn't a big deal.
It's also studying nurses early in the pandemic, who as a group were under unusual stress and exposed to unusual health risks, so not really generalisable to broader populations.