Lots of bugfixes and changes, but no new features.

Changes :
- It's no longer possible to accidentally select or move any of the image planes or axes or other critical Blender objects FRELLED needs - this can ONLY be done by the script. This means there's no longer any risk of slipping with the mouse and moving something important and seeing those funny square outlines all over the place. A very nice side-effect of this is that now it's easy to right-click to select mask regions in the 3D view (previously the image planes tended to get in the way).
- Changed the way in which the mbspect routine sets the masked spectral ranges. Previously it would mask the velocity range of any other source whose region box intersected the inspected source, even if it did so only slightly. In dense fields this could cause it to mask far too much of the target spectrum, since the boxes are generally quite a bit bigger than the sources. So the new approach is only to mask other sources if their box centre falls within 5 pixels of the target centre. This has a noticeable impact : most sources can now have their mbspect parameters determined perfectly well from FRELLED without any need to alter them by hand later.
- If you try and select a region that doesn't exist, everything will be unselected and a pop-up message will insult you. Previously absolutely nothing would happen, so there was a risk you could keep analysing the same source if you had one selected.
- The spectra shown on the screen with msbpect now has a comment showing the source name. This makes it easier to spot if there's a fatal error (which does not close the plotting window, so the previous spectrum is still displayed). Non-fatal errors still display the spectrum with a comment if there's a position-fitting failure.
- Regions which are set to wireframe mode are now always hidden from view in 2D mode unless the viewing plane intersects them, e.g. if the velocity channel overlaps their velocity range. Previously you could still see them unless the plane was "in front" of them, which was confusing and silly.
A side-effect of this is that objects that should not be visible are now automatically unselected by the background scripts, so if you pan through the cube and back again you'll lose your selection. This is because you can't unselect objects with the "A" key if they're hidden in this way, so if you wanted to unselect the object you'd have to pan back to it first. It seems more natural to me that you'd want them unselected so you can query NED and the SDSS at the cursor position.
- Changed how masks are imported. They are now first imported to a new blank scene and then linked to the other scenes. This is much faster than loading them directly into either the 3D scene (where there are hundreds of pre-made objects, making it slow to work with) or the 2D scene (where the frame has to be continually updated every time an object is loaded, slowing it down).
- Slight but quite nice speed increase when loading cubes, especially small ones. Delays were caused by creating placeholder objects (which can take significant time in files with hundreds of objects, so these are now included in the file by default) and a totally artificial 10 second wait statement (presumably left over from a long-forgotten debugging procedure).
- If you load a file in 2D mode, it will now automatically switch the view to one of the major projections (previously it would just display the axes, so you could be left wondering why you weren't seeing any data until you used the projection buttons in the Analyse script). If you only load one projection it will obviously orient the view for that projection. If you load two or three it will prefer XY first and XZ as a second choice.
- SDSS queries now use DR13 instead of DR10.

Bugfixes :
- Deleting objects previously only deleted them from the currently-visible scene (e.g. 3D or 2D mode). This meant that in some cases they didn't get properly deleted from the whole file and could cause problems. Now the objects selected in the currently visible scene are deleted from ALL scenes. Note that objects selected in one scene aren't necessarily selected in other scenes (so if you have different objects selected in different scenes and press delete, it's the ones selected in the current scene that get deleted - but they get completely deleted from the entire file).
- Masks imported with WCS coordinates will now display correctly in the RA-Dec projection if the velocity channel width in the FITS header is negative. This previously caused the mask objects to have a negative scale, so the scripts weren't properly evaluating if they should be visible as their channel range was not determined correctly.
- Fixed a problem where the script would crash if using simulation data with a partially-written FITS header. This happened because it would only identify data as simulated if the keywords identifying the coordinate system were missing. Now it also assumes the data is simulated if the keywords exist but specify anything other than equatorial or galactic coordinates (presumably no theoretician is malevolent enough to simulate data in equatorial or galactic coordinates...).
- Fixed bug in the 2D display that when importing masks before loading the data the highlighted grey channel range (the slider under the main window) was incorrect.
- The WCS mask file and html files now have each entry on separate lines, so they're easier to read directly if you need to for some reason.
- Removed an unnecessary print statement so long series of "Image_XXX_XXX" won't be printed to the terminal.
- Fixed a minor problem where the BMIN keyword was written to the FITS header text file as BMAJ (this doesn't seem to have mattered because the other scripts read the header by line, not by keywords).