Replace 'restr_label' with 'atom_site_id' in label imports

[vaitkus]

The following data items import the restr_label save frame:

• _restr_angle.atom_site_label_*
• _restr_distance.atom_site_label_*
• _restr_distance_min.atom_site_label_*
• _restr_equal_angle.atom_site_label_*
• _restr_equal_distance.atom_site_label_*
• _restr_equal_torsion.atom_site_label_*
• _restr_torsion.atom_site_label_*
• _restr_u_rigid.atom_site_label_*
• _restr_u_similar.atom_site_label_*

The human-readable description of restr_label from templ_attr.cif file reads:

      Labels of atom sites subtending distance or angle. Atom 2 is the apex for
      angular restraints.

Probably it is good enough, although I am unsure if it covers the _restr_u_* data items.

However, the same templ_attr.cif file contains a more generalised description of the atom label in save frame atom_site_id. The description of this save frame reads:

     This label is a unique identifier for a particular site in the
     asymmetric unit of the crystal unit cell.

This description does not provide information on atom 2 being the apex, however, this information can (and currently is) provided in the definition of the RESTR_ANGLE category. All other attributes of these save frames are identical/equivalent (the Encode purpose should be changed to Link in restr_label). As such, I suggest we switch from restr_label to atom_site_id and remove restr_label from templ_attr.cif altogether.

@jamesrhester, @nautolycus is this ok, or is there a reason to keep these definitions separate?

[jamesrhester]

The human-readable definition is important, and the intention here was clearly to try to describe the role the specific label played in describing the angle etc. However, any attempt at generality immediately fails as each atom plays a specific role as described in the category descriptions. So I agree that we can retire restr_label.