[llvm] [AMDGPU] Fix FinalizeISel skipping blocks after insertSimulatedTrap (PR #174774)

[Matt Arsenault via llvm-commits]

================
@@ -1981,8 +1982,8 @@ MachineBasicBlock *SIInstrInfo::insertSimulatedTrap(MachineRegisterInfo &MRI,
     MBB.addSuccessor(TrapBB);
   } else {
     // Since we're adding HaltLoopBB and modifying the CFG, we must return a
-    // different block to signal the change.
-    ContBB = HaltLoopBB;
+    // next block to signal the change.
----------------
arsenm wrote:

Can you fix it on the FinalizeISel side? This is a very implementation detail specific thing for the target uses to consider 

https://github.com/llvm/llvm-project/pull/174774