Support for D3(CSO) damping function

[awvwgk]

Add support for the CSO damping scheme proposed in https://pubs.acs.org/doi/10.1021/acs.jctc.5b00400.

Four parameter form

$$E_\text{disp} = - \sum_\text{AB}\left(s_6 + \frac{a_1}{1+\exp[R_\text{AB} - a_2 R_0^\text{AB}]}\right)\times \frac{C_6}{R^6_\text{AB} + (a_3 R_0^\text{AB} + a_4)^6}$$

One parameter form

$$E_\text{disp} = - \sum_\text{AB}\left(s_6 + \frac{a_1}{1+\exp[R_\text{AB} - 2.5 R_0^\text{AB}]}\right)\times \frac{C_6}{R^6_\text{AB} + (2.5^2)^6}$$

Published parameters

functional	s6	a1
BLYP	1.0	1.28
BP86	1.0	1.01
PBE	1.0	0.24
TPSS	1.0	0.72
B3LYP	1.0	0.86
PBE0	1.0	0.20
PW6B95	1.0	-0.15
B2PLYP	0.73	0.24

Potential issues:

• only eight functionals are parametrized with this scheme
• unclear how D3(CSO)-ATM would work