[llvm] [AMDGPU] Set Size to 4 for V_MOV_B64_PSEUDO and S_MOV_B64_IMM_PSEUDO (PR #70376)

[Stanislav Mekhanoshin via llvm-commits]

================
@@ -132,7 +132,7 @@ def V_MOV_B64_PSEUDO : VPseudoInstSI <(outs VReg_64:$vdst),
   let isAsCheapAsAMove = 1;
   let isMoveImm = 1;
   let SchedRW = [Write64Bit];
-  let Size = 16; // Needs maximum 2 v_mov_b32 instructions 8 byte long each.
+  let Size = 4;
----------------
rampitec wrote:

By the time we run BranchRelaxation we already have no these post-RA expandable pseudos. I.e. it should really return a correct size, but only at certain point of lowering.

https://github.com/llvm/llvm-project/pull/70376