hess and ohess don't support diatomic molecules?

[longdragoon]

Describe the bug
Hi I'm constructing a small molecule library including the H2 and HCl. The hess (or ohess, as well as the thermo submodule) stops working for these diatomic molecules. xtb can output the hessian and the vibspectrum file correctly but does not give the thermodynamic correction.

To Reproduce
xtb installed with conda from conda-forge
run with "xtb --gfn 1 xtbopt.xyz --hess --verbose >> xtbopt.out 2>> xtbopt.err"
h2.zip
stdout stopped after detecting this molecule has an Dinfh symmetry. stderr does not have any useful information.
all input and output files provided

Expected behaviour
Please support the thermodynamic corrections for linear molecules

[haneug]

I could not reproduce this with the downloadable binary but only with the conda version. I submitted a PR that should hopefully resolve this issue but I really do not know much about the conda build.

[longdragoon]

I could not reproduce this with the downloadable binary but only with the conda version. I submitted a PR that should hopefully resolve this issue but I really do not know much about the conda build.

Yes That's what I noticed, too.
Additionally, I think the two variables magic4 and magic5 (also conv4 and conv5) for calculating the translational entropy (s_tr) is not very correct in xtb/src/thermo.f90