MaybeUninit::assume_init_read should have noundef load metadata

I was looking into `array::IntoIter` optimization, and noticed that it wasn't annotating the loads with `noundef` for simple things like `array::IntoIter<i32, N>`.

Turned out to be a more general problem as `MaybeUninit::assume_init_read` isn't marking the load as initialized (<https://rust.godbolt.org/z/Mxd8TPTnv>), which is unfortunate since that's basically its reason to exist.

This PR lowers `ptr::read(p)` to `copy *p` in MIR, which fortuitiously also improves the IR we give to LLVM for things like `mem::replace`.

compiler/rustc_hir_analysis/src/check/intrinsic.rs

@@ -363,6 +363,8 @@ pub fn check_intrinsic_type(tcx: TyCtxt<'_>, it: &hir::ForeignItem<'_>) {
             sym::likely => (0, vec![tcx.types.bool], tcx.types.bool),
             sym::unlikely => (0, vec![tcx.types.bool], tcx.types.bool),
 
+            sym::read_via_copy => (1, vec![tcx.mk_imm_ptr(param(0))], param(0)),
+
             sym::discriminant_value => {
                 let assoc_items = tcx.associated_item_def_ids(
                     tcx.require_lang_item(hir::LangItem::DiscriminantKind, None),