New lint: Unnecessary use of std::ptr::{copy,copy_nonoverlapping} with slices

During a Secure Code WG audit, I ran into the following unsafe code patterns, which can always be replaced with safe code with essentially no drawback or overhead:

• std::ptr::copy{,_nonoverlapping}(&src_slice[i] as *const T, &mut dst_slice[j] as *mut T, n)
  All non-UB uses on Copy types can be replaced with dst.slice[j..j+n].copy_from_slice(src_slice[i..i+n]).
  Moreover, if T isn't Copy, this is UB and should be replaced with dst.slice[j..j+n].clone_from_slice(src_slice[i..i+n])

• std::ptr::copy{,_nonoverlapping}(&s[i] as *const T, &mut s[j] as *mut T, n)
  Same idea, non-UB uses can be replaced with s.copy_within(i..i+n, j).
  If the ranges are non-overlapping, it might be faster to use slice::split_at_mut and copy_from_slice (resulting in a call to std::ptr::copy_nonoverlapping), but that might be too much static analysis to ask from Clippy (though it can be safely assumed in the copy_nonoverlapping case)

No source file, test, or entry point is named. Start by reviewing how existing Clippy lints detect unsafe std::ptr::copy and copy_nonoverlapping patterns, then determine how the proposed slice replacements should be validated for Copy types, overlap, and ranges. Done means the lint reliably identifies the described cases and has coverage for its suggested replacements.