[mlir][emitc] Fix bug in ApplyOp translation (#155171)

The translator emits `emitc.apply` incorrectly when the op is part of an
expression, as it prints the name of the operand instead of calling
emitOperand() which takes into account the expression being emitted,
leaving out the part of the expression feeding this op, e.g.
```mlir
func.func @foo(%a: i32, %p: !emitc.ptr<i32>) -> i32 {
  %c = emitc.expression : i32 {
    %e = emitc.sub %p, %a : (!emitc.ptr<i32>, i32) -> !emitc.ptr<i32>
    %d = emitc.apply "*"(%e) : (!emitc.ptr<i32>) -> i32
    emitc.yield %d : i32
  }
  return %c : i32
}
```
translates to:
```C
int32_t foo(int32_t v1, int32_t* v2) {
  int32_t v3 = *v4;
  return v3;
}
```
instead of:
```C
int32_t foo(int32_t v1, int32_t* v2) {
  int32_t v3 = *(v2 - v1);
  return v3;
}
```

Author: aniragil

mlir/lib/Target/Cpp/TranslateToCpp.cpp

@@ -782,9 +782,7 @@ static LogicalResult printOperation(CppEmitter &emitter,
   if (failed(emitter.emitAssignPrefix(op)))
     return failure();
   os << applyOp.getApplicableOperator();
-  os << emitter.getOrCreateName(applyOp.getOperand());
-
-  return success();
+  return emitter.emitOperand(applyOp.getOperand());
 }
 
 static LogicalResult printOperation(CppEmitter &emitter,

mlir/test/Target/Cpp/expressions.mlir

@@ -295,6 +295,25 @@ func.func @different_expressions(%arg0: i32, %arg1: i32, %arg2: i32, %arg3: i32)
   return %v_load : i32
 }
 
+// CPP-DEFAULT:      int32_t expression_with_dereference(int32_t [[VAL_1:v[0-9]+]], int32_t* [[VAL_2]]) {
+// CPP-DEFAULT-NEXT:   int32_t [[VAL_3:v[0-9]+]] = *([[VAL_2]] - [[VAL_1]]);
+// CPP-DEFAULT-NEXT:   return [[VAL_3]];
+// CPP-DEFAULT-NEXT: }
+
+// CPP-DECLTOP:      int32_t expression_with_dereference(int32_t [[VAL_1:v[0-9]+]], int32_t* [[VAL_2]]) {
+// CPP-DECLTOP-NEXT:   int32_t [[VAL_3:v[0-9]+]];
+// CPP-DECLTOP-NEXT:   [[VAL_3]] = *([[VAL_2]] - [[VAL_1]]);
+// CPP-DECLTOP-NEXT:   return [[VAL_3]];
+// CPP-DECLTOP-NEXT: }
+func.func @expression_with_dereference(%arg1: i32, %arg2: !emitc.ptr<i32>) -> i32 {
+  %c = emitc.expression : i32 {
+    %e = emitc.sub %arg2, %arg1 : (!emitc.ptr<i32>, i32) -> !emitc.ptr<i32>
+    %d = emitc.apply "*"(%e) : (!emitc.ptr<i32>) -> i32
+    emitc.yield %d : i32
+  }
+  return %c : i32
+}
+
 // CPP-DEFAULT:      bool expression_with_address_taken(int32_t [[VAL_1:v[0-9]+]], int32_t [[VAL_2:v[0-9]+]], int32_t* [[VAL_3]]) {
 // CPP-DEFAULT-NEXT:   int32_t [[VAL_4:v[0-9]+]] = 42;
 // CPP-DEFAULT-NEXT:   return &[[VAL_4]] - [[VAL_2]] < [[VAL_3]];