In the two recent snapshot releases NWChem always reports rotational entropy as 0.0. The problem was a mistake made during refactoring src/vib/vib_wrtfreq.F to support single atoms. The inertprod value is now initialized to zero and never updated.
If you put
inertprod=inertia_A*inertia_B*inertia_C
back in before
if (dabs(inertprod).lt.1.0d-9) then
around line 238 of vib_wrtfreq.F, the code properly reports rotational entropy again.