CG and SQNM output now all work.
Found issue in CG with backtracking line-min where use of PR gave
gamma of zero consistently, but FR works; now default to FR if PR
gives gamma < 0
MD reworking underway
Introduce option to set derivatives at start and end of table
when reading radial tables. Specifically used to ensure PCC
charge derivative goes to zero at r=0.
The variable atomicnum from dimens_module.f90 is integer, so we
need to round it appopriately now that z is real; I have added
a call to nint to do this.
Debug: changed the type of z (member of pseudo_info) from integer to real,
following the change of the format of *.ion file.
(I have changed the type in MakeIonFiles and forgot to change it in CONQUEST.)