Corrections to NIST dataset

[gabrielasd]

This pull request addresses the following issues related to the NIST dataset:

• Incorrect energy conversion from $cm^{-1}$ to $Ha$ in the compilation script.
  For example the energy for hydrogen should be about $-0.5 Ha$ not this humongous value:

h_neutral = load(elem='H', charge=0, mult=2, dataset='nist')
print(h_neutral.energy)     # -24071707996.6 Ha ?!!!

this is caused by using $energy * CMINV$ when converting units. I've changed this to $energy / CMINV$ but alternatively we may want to redefine the CMINV variable.

• Anionic species with charge -1 returning None value of energy.
  This problem was raised in issue NIST dataset #100. As solution I used: $E_{anion} = E_{neutral} - IP_{anion}$

• Incorrect variable being returned by the Species.atmass attribute.
  This is not a specific problem of NIST dataset, but was caught by this dataset tests in PR Add zero-electron species to gaussian dataset #138 causing the checks to fail for python > v3.9. The attribute was returning the value from the .msg file instead of the one from the Element class instance in the periodic module.

This changes are a WIP, the following tasks remain to be completed:

• Fix NIST test module
• Recompile NIST test files