Start avoiding parse diagnostics on error tokens (#4431)

An invalid parse due to an error token isn't likely a great diagnostic
as it will already have been diagnosed by the lexer. A common case to
start handling that is when the parser encounters an invalid token when
expecting an expression.

This removes a number of unhelpful diagnostics after the lexer has done
a good job diagnosing.

This also means that there may be parse tree errors that aren't
diagnosed when there are lexer-diagnosed errors, so track that.

Follow-up to #4430 that almost finishes addressing its diagnostic TODO.

toolchain/parse/handle_expr.cpp

@@ -73,7 +73,7 @@ auto HandleExprInPostfix(Context& context) -> void {
   // Parses a primary expression, which is either a terminal portion of an
   // expression tree, such as an identifier or literal, or a parenthesized
   // expression.
-  switch (context.PositionKind()) {
+  switch (auto token_kind = context.PositionKind()) {
     case Lex::TokenKind::Identifier: {
       context.AddLeafNode(NodeKind::IdentifierNameExpr, context.Consume());
       context.PushState(state);
@@ -208,11 +208,15 @@ auto HandleExprInPostfix(Context& context) -> void {
       break;
     }
     default: {
+      // If not already diagnosed in the lexer, diagnose it here.
+      if (token_kind != Lex::TokenKind::Error) {
+        CARBON_DIAGNOSTIC(ExpectedExpr, Error, "expected expression");
+        context.emitter().Emit(*context.position(), ExpectedExpr);
+      }
+
       // Add a node to keep the parse tree balanced.
       context.AddLeafNode(NodeKind::InvalidParse, *context.position(),
                           /*has_error=*/true);
-      CARBON_DIAGNOSTIC(ExpectedExpr, Error, "expected expression");
-      context.emitter().Emit(*context.position(), ExpectedExpr);
       context.ReturnErrorOnState();
       break;
     }