mksurfdata_esmf should check that pfts 1-16 end up with hbot htop sai > 0 when lai > 0

In #3501 the test `ERP_D_Ld3_PS.f09_g17.I2000Clm50Sp.derecho_intel.clm-prescribed` that reads LAI via streams, uncovered a problem originating in the LAI raw data that was intended for ctsm5.4:
- The new LAI raw data (unlike prior LAI raw data) stops with pft 14 and does not include pft 15
- As a result, fsurdat files end up with lai, sai, htop, hbot = 0 for pfts 15 and 16
- Test-suite tests not reading LAI via streams do not fail because the model sends pfts 15 and 16 through BareGroundFluxes
- The "prescribed" test fails because lai > 0 sends pfts 15 and 16 through CanopyFluxes where sai, htop, hbot = 0 result in z0 = 0, which enters a log function and triggers a segmentation fault

This problem could have been caught in mksurfdata_esmf by checking that grid cells with lai > 0 end up with hbot htop sai > 0, else there's a problem.