We should calibrate or redefine the factor 2.0 in get_r_inner for response calculations

[bast]

This possible problem was pointed out by @jaimeaxel.

https://github.com/dftlibs/numgrid/blob/master/numgrid/AtomGrid.cpp#L94-L95

[bast]

The factor is hard-coded and may not be enough for response.

[bast]

numgrid/src/radial.rs

Lines 80 to 81 in 319a80e

	// factor 2.0 to match DIRAC code
	let r_inner = get_r_inner(radial_precision, alpha_max * 2.0);