Inline asm: canonicalizer does not respect `i` and `d` constraints
• Given the following LLVM MLIR dialect example (example.mlir ): llvm.func @f(%arg0: i16) { %0 = llvm.mlir.constant(64 : i16) : i16 %1 = llvm.mlir.constant(32 : i16) : i16 %5 = llv
Respect