if subroutine angle is the calling program, we always have
ntab1=2 and ntab2=3. the same is true if subroutine partl is the
calling program and partl was called by subroutine optic or trnsit
if subroutine deldop is the calling program, we usually have
ntab1=2 and ntab2=3, but the values ntab1=1 and ntab2=3 or ntab1=1
and ntab2=2 are possible.  if subroutine partl is the calling
program and partl was called by subroutine radar, then we usually
have ntab1=2 and ntab2=3, but ntab1=1 and ntab2=3 is also possible
only the values tabvl(i) for i=ntab1 to ntab2+8 are used in the
computations in subroutine ycoff.  the dimension of y in the
calling program is sometimes (7,3,3) and the calling statement
looks like;
call YCOFF(y(1,1,m))

note: the f vector could, in principle, be dotted with evc7(2:7,i),
but the magnitude of the coefficients drops off with increasing
index.  therefore, roundoff errors are minimized by adding the
terms in reverse order