[X86] Use X86ISD::MUL_IMM instead of ISD::MUL for multiply we intend to be selected to LEA.

This prevents other combines from possibly disturbing it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@337890 91177308-0d34-0410-b5e6-96231b3b80d8
1 file changed