TABLE_MARGINALIZE
This taks is used to project (marginalize) a table of N+1 columns, one
of which is contains the value of a function of the N others, onto a ta-
ble of 3 columns, containing the "marginalized" value of the function of
only 2 parameters, columns noted X,Y hereafter. The table is assumed to
be regularly spaced in (X,Y) values in this current version.
This process is often needed in multi-parameter minimization, e.g. to
derive a 2-D Chi^2 surface from an (N-1)-D hypersurface (or equivalent-
lyfor a Likelihood distribution).
To derive a single value from the N-2 values for each (X,Y) pair, a re-
duction operation is required. The available operations are MIN (appro-
priate for Chi^2 projection), MAX (appropriate for Likelihood) MEAN and
SUM.