This was my first reaction as well, but checking how I actually use a mouse I don't think it makes sense. Curious if it's jut me though.
For starters for normal desktop usage like >95% of the time the thing I want to click on isn't under the mouse yet so it doesn't obscure anything. Instead I move the mouse to it and by the time the mouse is there I don't care what is under it anymore because the decision to click it was made already.
Second when the mouse is over something I need to be able to read it seems I tend to move the mouse away (even when it's over text and turns into a straight cursor). The reason being that no matter what cursor is used its lines are typically wider and higher than the lines in rendered characters underneath so always obscures something. In other words: even if the cursor weren't tilted it would still obscure the same amount of surface, but just in a slightly different location. And that's really only slightly so for practical use won't matter.
Wrt where something is approached from: that depends on where the cursr is and where the target is. It would be really interesting to put this in a heatmap from daily usage, but I quickly checked some of the things I access often, like in my bottom taskbar and browser tab bar at the top, and since the things clicked often are both left and right of the screen and my cursor can be seeminlgy anywhere there might still be one approach direction used more but only by some marging.
Lastly most software I checked where obsucring could actually matter (e.g. CAD) uses custom cursors like 1 pixel wide crosses etc, not something tilted.