[mono] fix bound checks for Spans in LLVM-jit mode (add extra SROA pass) (#20054)

[mono] fix bound checks for Spans in LLVM-jit mode (add extra SROA pass) (#20054)

I noticed that the following code leaves bound checks when compiled using LLVM-JIT:
```csharp
static void Test(Span<int> span)
{
    for (int i = 0; i < span.Length; i++)
        x[i] = 0;
}
```
#### Current codegen:
```asm
Test:
0000000000000000	pushq	%rax
0000000000000001	testl	%esi, %esi
0000000000000003	jle	0x3f
0000000000000005	xorl	%eax, %eax
0000000000000007	movl	%esi, %ecx
0000000000000009	cmpq	%rcx, %rax
000000000000000c	jae	0x2c
000000000000000e	nop
0000000000000010	movl	$0x0, (%rdi)
0000000000000016	incq	%rax
0000000000000019	movslq	%esi, %rcx
000000000000001c	cmpq	%rcx, %rax
000000000000001f	jge	0x3f
0000000000000021	addq	$0x4, %rdi
0000000000000025	movl	%esi, %ecx
0000000000000027	cmpq	%rcx, %rax
000000000000002a	jb	0x10
000000000000002c	movabsq	$0x7fc330531710, %rax
0000000000000036	movl	$0x106, %edi
000000000000003b	callq	*(%rax) ;; <---- bound check
000000000000003d	ud2
000000000000003f	popq	%rax
0000000000000040	retq
```
looks like an additional -sroa pass after lower-expect and instcombine fixes it:
#### New codegen:
```asm
Test:
0000000000000000	testl	%esi, %esi
0000000000000002	jle	0x1f
0000000000000004	movl	%esi, %eax
0000000000000006	nopw	%cs:(%rax,%rax)
0000000000000010	movl	$0x0, (%rdi)
0000000000000016	addq	$0x4, %rdi
000000000000001a	decq	%rax
000000000000001d	jne	0x10
000000000000001f	retq
```
Here is the original LLVM IR before optimizations: https://godbolt.org/z/PLfR6x

Co-authored-by: EgorBo <EgorBo@users.noreply.github.com>