GXF and Imagine Fixes

Folks,

I have fixed the bug in GXF projections that N. Farah reported.  In doing
so I found a very serious bug in the GXF low level code with images that
are taller than they are wide ... basically memory corruption is likely.
Anyone depending on the GDAL GXF support should look at incorporating this
fix for a long standing bug (in gxfopen.c).  The projection bug was also
fixed but was unrelated.

The second issue is that I discovered I botched hfaband.cpp a few weeks
ago when incorporating better error reporting.  Basically, Erdas Imagine
(HFA) support has been completely broken since September 10th.

While bugs are going into GDAL and being fixed on a regular basis, I will
try to warn of serious ones I discover.

Best regards,

--
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam@xxxxxxxxx
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent