As suggested by others the self-test should only ever be run when the board is cold, with nothing attached to the RF ports and probably ideally not fitted in a case — impossible to say how pressure on the board and adjacent metalwork etc. might affect results.
Also, hopefully it’s a one off, but we haven’t tested the Hacker Gadgets cases and obviously cannot vouch for third party products anyway, but it is possible that thermal coupling may not be great and if so, this might not help either.
Although you note that it does seem to be keeping the temperature down, which is good, and the other report regarding the case will hopefully be a one-off QC issue.