Strange Fermi energy for gapped states

[tsuyoshi38]

We usually assume that the Fermi energy (used in DOS plot) for gapped states is set as the middle of valence band top and conduction band bottom. But, it is not like that in some cases. See the attached input files provided by Timothee.
In this case, Fermi energy can be either the valence band top or the conduction band bottom.

I wonder lower and higher bounds in findFermi_varspin can be almost the same, in Fermi-Dirac distribution function depending on the value of tolElec. But, I have not tried it yet. (I also worry that the long tail in Fermi-Dirac function might cause a problem.)
Input_issue_Fermi_energy.zip