Avoid hash allocation for certain proc calls

Previously, proc calls such as:

```ruby
proc{|| }.(**empty_hash)
proc{|b: 1| }.(**r2k_array_with_empty_hash)
```

both allocated hashes unnecessarily, due to two separate code paths.

The first call goes through CALLER_SETUP_ARG/vm_caller_setup_keyword_hash,
and is simple to fix by not duping an empty keyword hash that will be
dropped.

The second case is more involved, in setup_parameters_complex, but is
fixed the exact same way as when the ruby2_keywords hash is not empty,
by flattening the rest array to the VM stack, ignoring the last
element (the empty keyword splat).  Add a flatten_rest_array static
function to handle this case.

Update test_allocation.rb to automatically convert the method call
allocation tests to proc allocation tests, at least for the calls
that can be converted.  With the code changes, all proc call
allocation tests pass, showing that proc calls and method calls
now allocate the same number of objects.

I've audited the allocation tests, and I believe that all of the low
hanging fruit has been collected.  All remaining allocations are
either caller side:

* Positional splat + post argument
* Multiple positional splats
* Literal keywords + keyword splat
* Multiple keyword splats

Or callee side:

* Positional splat parameter
* Keyword splat parameter
* Keyword to positional argument conversion for methods that don't accept keywords
* ruby2_keywords method called with keywords

Reapplies abc04e898b627ab37fa9dd5e330f239768778d8b, which was reverted at
d56470a27c5a8a2e7aee7a76cea445c2d29c0c59, with the addition of a bug fix and
test.

Fixes [Bug #20679]

1 files changed, 26 insertions, 13 deletions

diff --git a/vm_args.c b/vm_args.c
index 84ffd2e983..0ecdc31a31 100644
--- a/vm_args.c
+++ b/vm_args.c
@@ -571,6 +571,22 @@ check_kwrestarg(VALUE keyword_hash, unsigned int *kw_flag)
     }
 }
 
+static void
+flatten_rest_args(rb_execution_context_t * const ec, struct args_info *args, VALUE * const locals, unsigned int *ci_flag)
+{
+    const VALUE *argv = RARRAY_CONST_PTR(args->rest);
+    int j, i=args->argc, rest_len = RARRAY_LENINT(args->rest)-1;
+    args->argc += rest_len;
+    if (rest_len) {
+        CHECK_VM_STACK_OVERFLOW(ec->cfp, rest_len+1);
+        for (i, j=0; rest_len > 0; rest_len--, i++, j++) {
+            locals[i] = argv[j];
+        }
+    }
+    args->rest = Qfalse;
+    *ci_flag &= ~VM_CALL_ARGS_SPLAT;
+}
+
 static int
 setup_parameters_complex(rb_execution_context_t * const ec, const rb_iseq_t * const iseq,
                          struct rb_calling_info *const calling,
@@ -727,35 +743,32 @@ setup_parameters_complex(rb_execution_context_t * const ec, const rb_iseq_t * co
                 args->rest_dupped = false;
 
                 if (ignore_keyword_hash_p(rest_last, iseq, &kw_flag, &converted_keyword_hash)) {
-                    if (ISEQ_BODY(iseq)->param.flags.has_rest || arg_setup_type != arg_setup_method) {
+                    if (ISEQ_BODY(iseq)->param.flags.has_rest) {
                         // Only duplicate/modify splat array if it will be used
                         arg_rest_dup(args);
                         rb_ary_pop(args->rest);
                     }
+                    else if (arg_setup_type == arg_setup_block && !ISEQ_BODY(iseq)->param.flags.has_kwrest) {
+                        // Avoid hash allocation for empty hashes
+                        // Copy rest elements except empty keyword hash directly to VM stack
+                        flatten_rest_args(ec, args, locals, &ci_flag);
+                        keyword_hash = Qnil;
+                        kw_flag = 0;
+                    }
                     given_argc--;
                 }
                 else if (!ISEQ_BODY(iseq)->param.flags.has_rest) {
                     // Avoid duping rest when not necessary
                     // Copy rest elements and converted keyword hash directly to VM stack
-                    const VALUE *argv = RARRAY_CONST_PTR(args->rest);
-                    int j, i=args->argc, rest_len = RARRAY_LENINT(args->rest)-1;
-                    args->argc += rest_len;
-                    if (rest_len) {
-                        CHECK_VM_STACK_OVERFLOW(ec->cfp, rest_len+1);
-                        for (i, j=0; rest_len > 0; rest_len--, i++, j++) {
-                            locals[i] = argv[j];
-                        }
-                    }
-                    args->rest = Qfalse;
-                    ci_flag &= ~VM_CALL_ARGS_SPLAT;
+                    flatten_rest_args(ec, args, locals, &ci_flag);
 
                     if (ISEQ_BODY(iseq)->param.flags.has_kw || ISEQ_BODY(iseq)->param.flags.has_kwrest) {
                         given_argc--;
                         keyword_hash = converted_keyword_hash;
                     }
                     else {
+                        locals[args->argc] = converted_keyword_hash;
                         args->argc += 1;
-                        locals[i] = converted_keyword_hash;
                         keyword_hash = Qnil;
                         kw_flag = 0;
                     }