* generation of noncoliner magnetization from lsda magnetization moved out of
  the routine reading rho into "potinit"
* charge density with noncolinear magnetization so obtained is written to file
  at the end of the noncolinear calculation 
More logical IMHO, no need to tweak low-level I/O routines that can be reverted
to previous state, postprocessing codes read the correct magnetization

Case pw_readfile('rho') deleted: it is useless and confusing (the entire 
routine is a useless wrapper that must disappear, but this another time)


git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@12170 c92efa57-630b-4861-b058-cf58834340f0