Fix NMR shieldings, closes #384

[stigrj]

The smoothing parameter for the r/r^3 function, as computed for use in nuclear gradients, was not sufficiently tight for NMR calculations. The old parameter was computed as described by Yanai et.al. as

c = [prec / (0.00435*Z^5*N_atoms)]^(1/3)
  = [prec / 0.00435]^(1/3), with Z = N_atoms = 1 in the NMR operator

The new parameter is simply set to

c = prec, in the NMR operator

This is the same as in v1.0.1, and seems to give nice and consistent NMR parameters, but we should reconsider the forms and parameters of all our smoothed functions (nuclear potential, nuclear gradient, NMR operator, ZORA potential, etc). A quick peek into the MADNESS source reveals that they are no longer using the same form for their potentials.

[codecov[bot]]

Codecov Report

Merging #385 (5c67294) into master (799534c) will decrease coverage by 0.00%. The diff coverage is 19.23%.

@@            Coverage Diff             @@
##           master     #385      +/-   ##
==========================================
- Coverage   69.95%   69.95%   -0.01%     
==========================================
  Files         177      177              
  Lines       14111    14110       -1     
==========================================
- Hits         9872     9871       -1     
  Misses       4239     4239              

Impacted Files	Coverage Δ
src/qmoperators/one_electron/H_BM_dia.h	0.00% <0.00%> (ø)
src/qmoperators/one_electron/H_MB_dia.h	0.00% <0.00%> (ø)
src/driver.cpp	70.16% <40.00%> (+0.04%)	:arrow_up:
src/analyticfunctions/NuclearGradientFunction.h	100.00% <100.00%> (ø)
...qmoperators/one_electron/NuclearGradientOperator.h	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 799534c...5c67294. Read the comment docs.