Track the start of a signature more accurately (#6760)

This change ensures that a function signature always starts with an
`IdentifierNameMaybeBeforeSignature` node (renamed from
`IdentifierNameBeforeParams`), even in the case of function declarations
like `fn F -> T` that have no parameter list. As a consequence, this
ensures that we push new entries onto `pattern_block_stack` and
`full_pattern_stack` when we start processing the function signature.

toolchain/check/handle_function.cpp

@@ -52,19 +52,6 @@ auto HandleParseNode(Context& context, Parse::FunctionIntroducerId node_id)
 auto HandleParseNode(Context& context, Parse::ReturnTypeId node_id) -> bool {
   auto [type_node_id, type_inst_id] = context.node_stack().PopExprWithNodeId();
 
-  // If the previous node was `IdentifierNameBeforeParams`, then it would have
-  // caused these entries to be pushed to the pattern stacks. But it's possible
-  // to have a fn declaration without any parameters, in which case we find
-  // `IdentifierNameNotBeforeParams` on the node stack. Then these entries are
-  // not on the pattern stacks yet. They are only needed in that case if we have
-  // a return type, which we now know that we do.
-  if (context.node_stack().PeekNodeKind() ==
-      Parse::NodeKind::IdentifierNameNotBeforeParams) {
-    context.pattern_block_stack().Push();
-    context.full_pattern_stack().PushFullPattern(
-        FullPatternStack::Kind::ExplicitParamList);
-  }
-
   // Propagate the type expression.
   auto form_expr = ExprAsReturnForm(context, type_node_id, type_inst_id);
   context.PushReturnForm(form_expr);