By the way, last time I was playing with raycasts intensively, I had to position them and wait one frame before I could read their result. That makes sense. But If I have one raycast and I have to trie every direction (let say 32 directions). I'll have to wait 32 frames before I checked every direction. Let say we optimize the process and we use an area sphere, like I did already in my 2nd attempt, so we can pinpoint the raycast to every detected object. If there are many objects in the area, the loss of time is proportional to the number of objects.
This is why I'm quite cautious about raycasts.