rHEALPix projection combine caps bug

[spethm]

Projecting from lon/lat using the rHEALPix projection produces incorrect projections if the polar regions are shifted and the on/lat are in the polar regions. If npole or spole are in the range of [1,3] inclusive.

More detailed inforamtion on the bug including graphs can be found here:  https://df.bestgrid.org/quickshare/5c677519931baa0e/SCENZ-CombineCaps-2012-06-07.pdf

[spethm]

This bug has been fixed with the following patch:  https://df.bestgrid.org/quickshare/cae33e5e7caf3955/combine_caps_bug-2012-06-06.patch

HEALPix and rHEALPix projection testing documentation can be found here (tested with the patched version of proj4):  https://df.bestgrid.org/quickshare/d6b285865b34db70/SCENZ-ProjectionTests-2012-06-07.pdf

Binaries used in the testing for the combine caps bug as well as overall projection testing:  https://df.bestgrid.org/quickshare/7364e5e560b6d60d/jproj4-test.tar.bz2

Source for the tests:  https://df.bestgrid.org/quickshare/3ad729a5bf29e17a/jproj4-test-src.tar.bz2

[warmerdam]

I have incorporated the suggested patch (r2226). While I reviewed the .pdf docs briefly I was not easily able to deduce a appropriate test point to check the fix. If you can suggest an addition to nad/testvarious I'd be pleased to add it.

[spethm]

I think I can add a few tests to nad/testvarious that incorporates the polar caps at positions other then the default which is 0. I'll work on that and submit a patch.

[spethm]

I have added new tests to the testvarious script to checks for a single point on the north polar region and the south polar region across all possible polar region positions.

The patch can be found here:  https://df.bestgrid.org/quickshare/2d5799f18b11a37e/combe_caps_test-2012-07-18.patch