[TableGen][CodeEmitterGen] Fix SubOpAliases MIOperandNo mixup

[iii]

Summary

SubOpAliases maps a sub-operand name to the respective operand's index
and the sub-operand number within this operand. The operand index is
used for the Operands array.

Currently MIOperandNo is used as the operand index, which is not
correct. For example, if there are 2 operands with 3 sub-operands each:

(ins (bdladdr12onlylen4 $B1, $D1, $L1):$BDL1,
     (bdladdr12onlylen4 $B2, $D2, $L2):$BDL2)

then B2's operand index will be 3, but the correct value is 1.

[jyknight]

Huh, whoops -- does this change any tablegen codegen that's in-tree now?

[iii]

Fortunately not. I found this while adding named sub-operand support to SystemZ.

[jyknight]

Great, thanks!

llvm/utils/TableGen/CodeGenInstruction.cpp

if (SubArgDag) {
         if (auto MaybeEncoderMethod =
                 cast<DefInit>(MIOpInfo->getArg(j))
                     ->getDef()
                     ->getValueAsOptionalString("EncoderMethod")) {
           OpInfo.EncoderMethodNames[j] = *MaybeEncoderMethod;
         }
 
         OpInfo.SubOpNames[j] = SubArgName;
-        SubOpAliases[SubArgName] = std::make_pair(MIOperandNo, j);
+        SubOpAliases[SubArgName] = std::make_pair(i, j);
       }
     } else if (!EncoderMethod.empty()) {
       // If we have no explicit sub-op dag, but have an top-level encoder
       // method, the single encoder will multiple sub-ops, itself.
       OpInfo.EncoderMethodNames[0] = EncoderMethod;
       for (unsigned j = 1; j < NumOps; ++j)
         OpInfo.DoNotEncode[j] = true;
     }