Many of the files had some sort of optical problem.  I threw a bunch of
them out but kept a few of the marginal cases.  In the second pass the 
marginal cases all looked good, so part of the "optical noise" may have been
a bad starting rgm.  The 11 files below are the ones I used.  The difference
between rgm_strict.txt and rgm_relax.txt was only 3 pixels, so I opted to
use rgm_strict.txt as the official rgm.

sharc2-055349.fits
sharc2-055353.fits
sharc2-055355.fits
sharc2-055405.fits
sharc2-055440.fits
sharc2-055441.fits
sharc2-055472.fits
sharc2-055474.fits
sharc2-055514.fits
sharc2-055515.fits
sharc2-055534.fits