Correct Polar Alignment issue for telescopes with positive parity.

This fixes a long-standing bug in the KStars WCS implementation, but one which really only affected polar alignment.

The issue relates to imaging systems where, if the image were rotated so that up would be North, then East would be to the left (lower X pixel coordinate). Most users have systems where East would be to the right. However, users with reflector telescopes in prime focus (e.g. one mirror, no secondary, as in a RASA) have their images flipped as described.

The solvers actually know if the image is flipped this way, but that information was not being passed into the WCS methods. The FITSData methods wcsToPixel() and pixelToWCS() thus would only be correct, for flipped images, in the center pixel. However, that's all that was used outside of Polar Alignment. That's likely why this was not widely noticed.

The correction is simple. Pass the image parity information to the WCS setup method, FITSData::injectWCS(), and in there negate the CDELT1 FITS header value.

For this to properly detect flipped images with ASTAP and "local astrometry", a version of StellarSolver created on or after Wed Apr 7 must be used. Using the "internal solver" should work with previous StellarSolver versions.