Add some more accessible note that the first 20 zeros in env are important.

[MarcusTL12]

It is of course hinted to in the examples and in code using libcint (Fermi.jl, pyscf), but currently the only place I have found an explanation for why those are needed is in issue #70. Having a note about it in the readme or something would help a bit when figuring out how to interface directly with the library.

[davidsousarj]

@MarcusTL12 I am trying to implement the integrals from libcint in my program (my program is in fortran and I don't know much about C) and indeed I didn't understand why the pointer of the example starts at 20. I wrote my program with the pointer starting at 1 (fortran indexes are 1-based) and it worked fine so far, at least for the overlap integrals.

I wonder if it would be a problem for my case (I intend to use only cartesian, overlap, kinetic, nuclear and 2e integrals).

[MarcusTL12]

@davidsousarj We figured out that there is a bit for documentation for it in the cint.h.in file. The first 20 numbers of env are reserved for global variables which are explained under the // global parameters in env section in the file. The first number specifies the integral cutoff threshold, the next three specify the coordinate origin that is used some of the dipole/quadrupole type integrals. You should really have the first 20 numbers free (and set to zero by default) so that you do not run into weird unexplanable problems down the road.

Hope this helps.