Fix an error for Lrama::Grammar::ParameterizingRule::Rhs#resolve_user_code when multiple execute method

[ydah]

  1) integration user defined parameterizing rules prints messages corresponding to rules
     Failure/Error: expect(out).to eq(expected)

       expected: "(2, 3)\n(2, 3)\n(-2, -1)\npair even odd: 5\n(1, 0)\n(1, 0)\n(-2, -1)\npair odd even: 1\n"
            got: "(2, 3)\n(3, 3)\n(-2, -1)\npair even odd: 5\n(1, 0)\n(0, 0)\n(-2, -1)\npair odd even: 1\n"

       (compared using ==)

       Diff:

       @@ -1,9 +1,9 @@
        (2, 3)
       -(2, 3)
       +(3, 3)
        (-2, -1)
        pair even odd: 5
        (1, 0)
       -(1, 0)
       +(0, 0)
        (-2, -1)
        pair odd even: 1

     # ./spec/lrama/integration_spec.rb:46:in `test_parser'
     # ./spec/lrama/integration_spec.rb:109:in `block (3 levels) in <top (required)>'

[ydah]

need: #410

[yui-knk]

[note] The same parameterizing rule can be instantiated multiple times. For example in "spec/fixtures/integration/user_defined_parameterizing_rules.y", pair is instantiated two times. Before this change, original user_code reference name is overwritten by the first instantiation and it's not overwritten by the second instantiation. Therefore we need to dup the user code.

[yui-knk]

Is it inevitable to depend on #410? It seems that we can reproduce the issue by these diff without [alias].
In general, if a patch depends on other(s), it will take a time to merge the patch because other PR(s) may need time to discussion.

diff --git a/spec/fixtures/integration/user_defined_parameterizing_rules.y b/spec/fixtures/integration/user_defined_parameterizing_rules.y
index 9dda3b2..308c7e3 100644
--- a/spec/fixtures/integration/user_defined_parameterizing_rules.y
+++ b/spec/fixtures/integration/user_defined_parameterizing_rules.y
@@ -25,6 +25,7 @@ static int yyerror(YYLTYPE *loc, const char *str);
                     {
                         $$ = $1 + $2;
                         printf("(%d, %d)\n", $1, $2);
+                        printf("(%d, %d)\n", $X, $2);
                         printf("(%d, %d)\n", $:1, $:2);
                     }
                 ;
diff --git a/spec/lrama/integration_spec.rb b/spec/lrama/integration_spec.rb
index fd87547..99af41f 100644
--- a/spec/lrama/integration_spec.rb
+++ b/spec/lrama/integration_spec.rb
@@ -97,10 +97,12 @@ RSpec.describe "integration" do
   describe "user defined parameterizing rules" do
     it "prints messages corresponding to rules" do
       expected = <<~STR
+        (2, 3)
         (2, 3)
         (-2, -1)
         pair even odd: 5
         (1, 0)
+        (1, 0)
         (-2, -1)
         pair odd even: 1
       STR

  1) integration user defined parameterizing rules prints messages corresponding to rules
     Failure/Error: expect(out).to eq(expected)

       expected: "(2, 3)\n(2, 3)\n(-2, -1)\npair even odd: 5\n(1, 0)\n(1, 0)\n(-2, -1)\npair odd even: 1\n"
            got: "(2, 3)\n(3, 3)\n(-2, -1)\npair even odd: 5\n(1, 0)\n(0, 0)\n(-2, -1)\npair odd even: 1\n"

       (compared using ==)

       Diff:

       @@ -1,9 +1,9 @@
        (2, 3)
       -(2, 3)
       +(3, 3)
        (-2, -1)
        pair even odd: 5
        (1, 0)
       -(1, 0)
       +(0, 0)
        (-2, -1)
        pair odd even: 1

     # ./spec/lrama/integration_spec.rb:46:in `test_parser'
     # ./spec/lrama/integration_spec.rb:109:in `block (3 levels) in <top (required)>'

Finished in 7.68 seconds (files took 0.1706 seconds to load)
14 examples, 1 failure

[ydah]

Is it inevitable to depend on #410? It seems that we can reproduce the issue by these diff without [alias]. In general, if a patch depends on other(s), it will take a time to merge the patch because other PR(s) may need time to discussion.

Ah, indeed it is. sorry. I updated this PR.

[yui-knk]

Thank you!