Add ECCE_DEBUG_ATOM_COLOR, to measure the uniform-hue bug instead of guessing

CSLoadColorsCmd is where ECCE hands each atom colour to the scene graph.
With ECCE_DEBUG_ATOM_COLOR=1 it prints every one, which separates the two
possibilities that look identical on screen in issue #83: correct numbers
per element (oxygen 1,0,0 and hydrogen 1,1,1) mean the data and the lookup
are fine and the fault is below us in the GL or driver path, while green
numbers would make it ours.

Relevant because the hue is reported on the Wayland machine only, with the
same code and the same data on the X11 one -- which already points below
the application, and this confirms it rather than assuming it.

Worth knowing for whoever picks it up: SoWxRenderArea requests no GL
visual attributes at all.  The whole GLX_RGBA / depth / accum / stencil
block is commented out behind "@todo Work on visual stuff later", so the
canvas takes whatever default the platform hands it -- and that default is
exactly the sort of thing that differs between X11 and XWayland.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EFMUNbw2UpRPmgsqviAowb

Author: andy

src/viz/sgcommands/CSLoadColorsCmd.C

@@ -57,6 +57,18 @@ bool CSLoadColorsCmd::execute()
          colors->atomColor.set1Value(index, red, green, blue);
       }
 
+   //  ECCE_DEBUG_ATOM_COLOR=1 prints every colour ECCE hands the scene
+   //  graph.  For the uniform-green-atoms report (#83) this separates the
+   //  two possibilities that look identical on screen: if the numbers here
+   //  are correct per element (oxygen 1,0,0 and hydrogen 1,1,1) then the
+   //  data and the lookup are fine and the fault is below us, in the GL
+   //  or driver path; if they all come out green, it is ours.
+   if (getenv("ECCE_DEBUG_ATOM_COLOR")) {
+      printf("[ATOMCOLOR] index=%d  r=%.3f g=%.3f b=%.3f%s\n",
+             index, red, green, blue, colors ? "" : "   (NO COLOR NODE)");
+      fflush(stdout);
+   }
+
    return true;
 }