Something about high pass:
You can find something like frequency in pictures similar to sound. Low frequencies create the rough picture, high frequencies add more and more details and edges.
There is a low pass filter, too: Gaussian blur. It removes more and more edges and details and only lets the low frequencies pass that make the rough picture. If you blur a black pixel on white background, the result is more or less light gray: All in all the area around that pixel had that nearly white color.
The high pass filter is just the difference between the original picture and the low pass filtered version. a black pixel on white background can have a difference of up to -255 compared to the blurred version, a bigger area of the same color will have a difference of 0, and a white pixel on black background can be up to +255 brighter. But as normal layers cannot store values between -255 and +255, that values are mapped to 0 to 255, with 128 being the neutral value.
That's why the high pass looks mostly gray: Lots of areas in pictures have soft gradients or little difference in color, so there's also only little difference between the original and the blurred picture.
'Clarity' is just one use case for the filter. We've had a
nice thread about frequency separation lately, that shows that there's some more practical use for that filter, maybe it also gives a little more idea about high pass and low pass.