raycaster: fixed bug that caused wrong gradient computation when missing values are encountered

bug-description: instead of the isosurface of potential temperature during an MCAO it appeared like surface grid-cells were lifted upwards in the vertical z-direction in geographic areas where the isosurface was close the earth's surface fixes: 1.) uncomment the tests for missing values in "volume_hybrid_utils.glgsl". Note: I uncommented both sections of code (line 119-122 and 443-446) because it appears likely that this fixes the bug for both routines "sampleHybridSigmaVolumeAtPos_LUT" and the other one in lines above. However, this should be checked. 2.) account for missing values in gradient computation by fixes in: "volume_sample_utils.glsl" 3.) account for missing values when the ray traverses first a data volume, then a geographic area with missing values (e.g. a mountain), and then another data-volume by fixes in: "volume_raycaster.glsl". Note: maybe the same fix has to be done in other areas of the code?