Hnuc non-zero for single-level run but hard zero for 2-level run

[zingale]

Compile science/ecsn as:

make COMP=gnu USE_OMP=FALSE

then run a single-level test as:

mpiexec -n 16 ./Maestro3d.gnu.MPI.ex input_ecsn amr.max_level=0

Looking at plotfile 0, fextrema shows:

 Hnuc                        -3119672478.9           4679508718.3    

Now run with amr.max_level=1 and Hnuc is hard zero.

This seems like a bug.