Incorrect number of radical electrons calculated for metals

[greglandrum]

Thanks to @lpravda for pointing this out:

In [4]: m = Chem.MolFromSmiles('[Mn+2]')                                                                                     

In [5]: m.Debug()                                                                                                            
Atoms:
	0 25 Mn chg: 2  deg: 0 exp: 0 imp: 0 hyb: 4 arom?: 0 chi: 0 rad: 3
Bonds:

The algorithm used really only works for main group elements.

Since it's not generally possible to calculate this value correctly, a good workaround is probably to just assign numValenceElectrons % 2 for anything that has a default valence of -1

[jasondbiggs]

@greglandrum - is it expected that after this issue is fixed, the radical electrons no longer change when adding bonds?

For example, the scandium atom has one unpaired electron in all three of {"[Sc]", "[ScH]", "[ScH2]", "[ScH3]"}.