[flang][cuda] Define _CUDA only when preprocessor is enabled (#155913)

From the CUDA Fortran programming guide: 
> If CUDA Fortran is enabled in compilation, either by specifying -⁠cuda
on the command line, and pre-processing is enabled by either the
-⁠Mpreprocess compiler option or by using capital letters in the
filename extension (.CUF, .F90, etc.) then the _CUDA macro is defined.

Move the definition of `_CUDA` to the compiler invocation.

flang/lib/Frontend/CompilerInvocation.cpp

@@ -1754,6 +1754,11 @@ void CompilerInvocation::setDefaultPredefinitions() {
                                     fortranOptions.predefinitions);
   }
 
+  if (frontendOptions.features.IsEnabled(
+          Fortran::common::LanguageFeature::CUDA)) {
+    fortranOptions.predefinitions.emplace_back("_CUDA", "1");
+  }
+
   llvm::Triple targetTriple{llvm::Triple(this->targetOpts.triple)};
   if (targetTriple.isOSLinux()) {
     fortranOptions.predefinitions.emplace_back("__linux__", "1");

flang/lib/Parser/parsing.cpp

@@ -96,9 +96,6 @@ const SourceFile *Parsing::Prescan(const std::string &path, Options options) {
     prescanner.AddCompilerDirectiveSentinel("$cuf");
     prescanner.AddCompilerDirectiveSentinel("@cuf");
   }
-  if (options.features.IsEnabled(LanguageFeature::CUDA)) {
-    preprocessor_.Define("_CUDA", "1");
-  }
   ProvenanceRange range{allSources.AddIncludedFile(
       *sourceFile, ProvenanceRange{}, options.isModuleFile)};
   prescanner.Prescan(range);

flang/test/Parser/cuf-sanity-tree.CUF

@@ -1,4 +1,4 @@
-! RUN: %flang_fc1 -fdebug-dump-parse-tree %s 2>&1 | FileCheck %s
+! RUN: %flang_fc1 -fdebug-dump-parse-tree -x cuda %s 2>&1 | FileCheck %s
 include "cuf-sanity-common"
 !CHECK: Program -> ProgramUnit -> SubroutineSubprogram
 !CHECK: | SubroutineStmt

flang/test/Parser/cuf-sanity-unparse.CUF

@@ -1,4 +1,4 @@
-! RUN: %flang_fc1 -fdebug-unparse %s 2>&1 | FileCheck %s
+! RUN: %flang_fc1 -fdebug-unparse -x cuda %s 2>&1 | FileCheck %s
 include "cuf-sanity-common"
 !CHECK: SUBROUTINE atcuf
 !CHECK: END SUBROUTINE