Protocols: update dist_thr to be tighter

[bastonero]

The threshold for the hp.x calculation for reconstructing the 3x3x3 real space interactions should be tighter. This can be controlled by the dist_thr in the HP input file. Instead of 6e-4, we should use a more "standard" thr of 1e-5 (e.g. as spglib is using).