21, the arrow indicating bunched up highlights brought down above clipping is pointing at Bob's histogram with the manipulated tone-curve, not the OP's, which does not have any peak at the right that is lower than zero. A low-key image with noise or specular highlights will show a small amount of bright pixels in the histogram, but I suspect there is neither in the OP's so there are no pixels in the right 1/8th of the image.
The OP's images are darker on the dark end while the nose of the sheep or the stars being about the same brightness between the camera preview and the LR-raw-treatment. This could be histogram truncation on the dark end, but not the light end, which would make everything darker, not just truncate the darks as the sheep and star images show.
So far the OP has shown examples of the camera and LR treatment being different, asserted that the defaults are not the problem and complained about the messenger and tone of the message when this was suggested, and also claimed that LR and the initial preview are always the same. If this is really what has always been observed, then it would seem that the OP has merely forgotten to compute any previews during the import, so the intial thumbanils are the camera embedded previews for as long as it takes LR to compute it's own treatment preview, which gives enough time to observe the difference.
The OP should post an original raw using dropbox or similar large-file-hosting service to verify the settings and see if it behaves similarly on other systems to help determine whether the issue is LR-specific or system-specific.
If the exact same image performs differently on two different computers' LRs, then about the LR or system environments is different.
If multiple computers show the same differences between the camera-embedded JPG that is initially visible (or visible again in DPP) and the LR-treatment in the same images, then it is something the camera is doing to the embedded-JPG that you see before the three-dots go away.