--------------- Leap tests

These tests compare Leap output with that of previous versions
of the same runs, and when there is a difference, energies are
calculated and compared with the previous version.

When comparing Leap output with previous runs of Leap, normally 
there should be no differences at all, or at most very small 
differences in the last digits of decimal numbers, due to floating 
point math implementations on different machines. The big exception
is solvation, which can result in different numbers of waters being
added due to subtle differences in machine precision.

(There are also tests for comparing sander energies for molecules
built by Leap with those for the same molecules built by the
older Prep, Link, Edit, and Parm programs ('plep'); these are
more for developers than users wanting to validate compilation.
These are documented and run in the test/leap/plep/ subdirectory.)

The first tests duplicate the generation of residue libraries from
the old 'prep.in' format, comparing the results with the canonical
versions in $AMBERHOME/dat/leap/lib/. These tests are followed by
setups of various AMBER jobs; if the resulting prmtop,prmcrd files 
from one of these differs from the canonical output in the appropriate 
ff* subdirectory, an energy calculation is run to help assess the 
practical import of the difference.




