[Bug tree-optimization/89967] Inefficient code generation for vld2q_lane_u8 under aarch64

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89967

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
So the main thing that needs to be handled is:
```
  # .MEM_144 = VDEF <.MEM_143>
  g2b = D.23244;
  # VUSE <.MEM_144>
  _29 = g1b.val[0];
  g1v_73 = VIEW_CONVERT_EXPR<int32x4_t>(_29);
  # VUSE <.MEM_144>
  _30 = g1b.val[1];
  g2v_74 = VIEW_CONVERT_EXPR<int32x4_t>(_30);
  # VUSE <.MEM_144>
  _31 = g2b.val[0];
  g3v_75 = VIEW_CONVERT_EXPR<int32x4_t>(_31);
  # VUSE <.MEM_144>
  _32 = g2b.val[1];

```

_31 = D.23244.val[0];
_32 = D.23244.val[2];

I suspect after that, then there will less moves. But I am getting closer to
that point.

I am already down 2 mov.