A couple of thoughts:
Yeah, in the image that started this thread Matrix Metering would tend to overexpose the image due to all the darker water tones filling the frame. But it's not that you '... missed your under exposure by a click or two...' it's that you missed proper exposure. IOW, exposing properly for a scene like that isn't 'under exposure' it's still correct exposure it's just that the meter is heavily influenced by all that dark area and if you don't compensate will end up over exposing the image.
Remember that exposure meters work on the assumption that the entire world has 18% reflectivity (mid tone) and since the camera doesn't know what kind of scene you're filling the frame with it will tend to over exposure largely dark scenes and under expose largely bright scenes (e.g. snow or bright sand). One way to think about the exposure compensation dial when shooting in an automated exposure mode is you're telling the camera that this particular scene is darker or brighter than mid-tone so you get the correct exposure or close to it. When shooting in spot metering mode you're basically telling the camera that the relatively small area under the metering spot (selected focus point) is brighter or darker than mid-tone and by how much so when spot metering the Egret's body you're telling the camera that you'd like those areas to be captured around 2 stops brighter than mid tone by dialing in roughly 2 stops of positive exposure compensation.
When you shoot raw a small embedded jpeg is also created, that's actually the image you see when you do an image review in the field right on the camera. The histogram and any highlight warnings (blinkies) you look at in the field are actually based on this embedded jpeg. So yeah, there can be a bit of difference between the histogram and blinkies for the embedded jpeg and the actual raw data with the raw data typically having a bit more exposure latitude. IOW, you can often get by with slightly brighter tones without hard clipping them in the raw file than what your blinkies or histogram might indicate for the embedded jpeg. But those differences are typically small as in a third to maybe half a stop or so. Basically checking exposure after a test shot in camera does use the embedded jpeg so it won't perfectly show you what the raw file sees but it gets you very close but sometimes, I'll accept a tiny bit of white clipping (tiny blinkie areas) in the field knowing I have a bit more exposure latitude when I process the raw files.
And of course the big issue with clipping highlights is losing important detail like in the wings of your all white Egret. But some clipping of specular highlights in areas without important detail like sparkles off of glass surfaces or sometimes small blown out tips of whitewater on a river or other things can be acceptable and help the image. So it's not like every clipped highlight kills an image, it's the clipping of areas that should retain detail that we're generally trying to avoid.