Also guys I heard many engineers say gain before feedback is independent of mic's sensitivity. But when I use Shure sm58 (cardiod) I found that GBF is more than when I DPA Gooseneck (hyper cardiod) why is that I don't understand.
I think you mean: "Gain before feedback is independent of the source" - different mics will absolutely have different points at which they feedback, feedback is dependent on the mics frequency response, the pickup pattern, and its location compared to the speakers (lets not forget the speakers pattern).
So in a given place in a room, a specific microphone will have a specific amount it can be gained up before feedback can occur, that could +4dB or it could be plus +50dB - whatever that level of gain that can be added it is independent of the person speaking or singing.
My main question is what should be the SPL level from main and delays in relation to each other.
While there are different methodologies how I typically set delays are:
1) I find the delay where the coverage of the delay starts, I place my mic there, and find the delay of the main - that is your time to set the delay speaker to.
2) After I find that time alignment for most situations I often move the delay speakers delay 1-2ms back - this will make the focus on the main speaker at the front of the room, rather than the delay/fill near you.
3) For level I usually set it so that the fill adds clarity and so it is not super noticeable that "I am walking in front of a speaker" I like to keep delays as natural as possible - When I walk from the custody of the mains into the fill, if it is noticeably louder it is too loud. Note if you do the delay trick from step 2 you can have the fill be slightly louder than without as our brain and ears are more sensitive to timing than they are to volume.