Alternative Test set reflection sorting routines

[Gerard 'CD' Kleywegt]

use of thin shells has been around for a while
an option to select test reflexions this way has been
implemented in DATAMAN (available from uppsala or ccp4)
see  Structure 4, 897-904 (1996) and  Acta Cryst D52, 826-828
(1996)
results are not spectacularly better than straightforward
random selection, not even with ridiculously fat shells
--gerard