(j3.2006) Feature creep

[Van Snyder]

I've submitted two tiny late papers, both advocating tiny feature creeps
having to do with intrinsic assignment.

09-206 advocates removing the requirement that the rank of <expr> and
<variable> be the same if <variable> is polymorphic.  There seems not to
be a technical reason to have required it.

09-207 advocates removing the runtime requirement that the rank of
<expr> and <variable> be the same if <variable> is an unallocated
allocatable array.  The alternative proposed is to allocate <variable>
with all lower and upper bounds equal to 1 if <expr> is scalar.