My guess is the system is pressurized during operation. The cap on the t-stat housing should have a spring to hold the sealing surface down. A new cap rated at 10 to 13 psi is an inexpensive experiment.
Another FWIW... I don't recall if this issue has been addressed and that is that possibility of vacuum leaks.
You can use carb cleaner to hunt with. A spray on an area like intake gaskets etc. and listen for a change in RPM.