fixing rope pytest benchmark grad accumulation (#3743)

#3349 removed grad accumulation, but rope benchmark implementation needs
an update to get that working.

Reference implementation.
```
           Model  Batch-Size  Sequence-Length  ... Forward-Time(ms)  Backward-Kernels  Backward-Time(ms)
0  Llama-2-7b-hf                      2             4096  ...            0.166                 5              0.857
0  Llama-3-8B                         2             8192  ...            0.567                 5              1.433
0  mistralai/Mistral-Nemo-Base-2407   1             4096  ...            0.138                 6              0.166
0  Qwen/Qwen2.5-7B-Instruct           1             4096  ...            0.072                 8              0.397
0  microsoft/Phi-3.5-mini-instruct    1             8192  ...            0.236                 6              0.494
```
after l2_cache clear
```
                             Model  Batch-Size  Sequence-Length  ... Forward-Time(ms)  Backward-Kernels  Backward-Time(ms)
0  Llama-2-7b-hf                      2             4096  ...            0.166                 5              0.870
0  Llama-3-8B                         2             8192  ...            0.567                 5              1.444
0  mistralai/Mistral-Nemo-Base-2407   1             4096  ...            0.138                 6              0.192
0  Qwen/Qwen2.5-7B-Instruct           1             4096  ...            0.072                 8              0.417
0  microsoft/Phi-3.5-mini-instruct    1             8192  ...            0.234                 6              0.516
```

Before this PR:
```
Name (time in us)                                                                       Mean                    Median
---------------------------------------------------------------------------------------------------------------------------------
test_rope_bwd_benchmark[executor='thunder'-variation='llama_2_7b_hf_rope']        1,192.8558 (14.56)        1,191.9040 (14.53)
test_rope_bwd_benchmark[executor='thunder'-variation='llama_3_8B_rope']           1,767.5348 (21.58)        1,766.8410 (21.54)
test_rope_bwd_benchmark[executor='thunder'-variation='hf_mistral_nemo_rope']        275.4680 (3.36)           275.7265 (3.36)
test_rope_bwd_benchmark[executor='thunder'-variation='hf_qwen2_rope']               488.4243 (5.96)           488.3105 (5.95)
test_rope_bwd_benchmark[executor='thunder'-variation='hf_phi3_rope']                757.9140 (9.25)           757.6910 (9.24)
---------------------------------------------------------------------------------------------------------------------------------
```

In this PR:
```
Name (time in us)                                                                    Mean                Median
-----------------------------------------------------------------------------------------------------------------------
test_rope_bwd_benchmark[executor='thunder'-variation='llama_2_7b_hf_rope']       871.5996 (5.23)       871.6050 (5.24)
test_rope_bwd_benchmark[executor='thunder'-variation='llama_3_8B_rope']        1,443.0095 (8.66)     1,442.9955 (8.67)
test_rope_bwd_benchmark[executor='thunder'-variation='hf_mistral_nemo_rope']     166.5515 (1.0)        166.4480 (1.0)
test_rope_bwd_benchmark[executor='thunder'-variation='hf_qwen2_rope']            386.4463 (2.32)       386.5565 (2.32)
test_rope_bwd_benchmark[executor='thunder'-variation='hf_phi3_rope']             452.3351 (2.72)       452.0685 (2.72)
-----------------------------------------------------------------------------------------------------------------------
```

With the existing issue on pytest/torch.profiler, if I instead run each
benchmark separately,
```
test_rope_bwd_benchmark[executor='thunder'-variation='llama_2_7b_hf_rope']     871.1912  871.2465
test_rope_bwd_benchmark[executor='thunder'-variation='llama_3_8B_rope']        1.4427  1.4427
test_rope_bwd_benchmark[executor='thunder'-variation='hf_mistral_nemo_rope']   191.6567  191.6795
test_rope_bwd_benchmark[executor='thunder'-variation='hf_qwen2_rope']          416.8007  416.8935
test_rope_bwd_benchmark[executor='thunder'-variation='hf_phi3_rope']           514.7512  514.4900
```

So these number does match the manual benchmark with l2_cache cleared. I
think that justifies this PR.

benchmarks/python/test_rope.py

@@ -31,14 +31,15 @@ def test_rope_fwd_benchmark(
     elif executor == "torchcompile":
         clear_dynamo_cache()
 
-    model, inputs, _, _ = rope_setup[variation]()
+    model, gen_inputs, _, _ = rope_setup[variation]()
+    inputs = gen_inputs()
 
     def fwd_call(inp):
         return model(*inp)
 
     # Compile the fwd fn for torchcompile
     benchmark_fn = with_executor(executor, fwd_call, **kwargs)
-    run_benchmark(benchmark, benchmark_fn, inputs())
+    run_benchmark(benchmark, benchmark_fn, inputs)
 
 
 @pytest.mark.parametrize(
@@ -65,14 +66,15 @@ def test_rope_bwd_benchmark(
     elif executor == "torchcompile":
         clear_dynamo_cache()
 
-    model, fwd_inputs, grad, iobytes = rope_setup[variation]()
+    model, gen_inputs, grad, iobytes = rope_setup[variation]()
+    fwd_inputs = gen_inputs()
 
     def fwd_call(inp):
         return model(*inp)
 
     # execute the compiled fwd fn
     fwd_fn = with_executor(executor, fwd_call, **kwargs)
-    outputs = fwd_fn(fwd_inputs())
+    outputs = fwd_fn(fwd_inputs)
 
     # accumulate all output, so we can feed a single grad and use the unary bwd function
     output = outputs[0]
@@ -82,5 +84,5 @@ def fwd_call(inp):
     # NOTE: the iobytes is computed based on how thunder autograd worked. So this is just
     # a reference point for torchcompile and eager executor for comparison.
     run_benchmark(
-        benchmark, unary_bwd_torch, [output, grad(), fwd_inputs()], iobytes=iobytes()
+        benchmark, unary_bwd_torch, [output, grad(), *fwd_inputs], iobytes=iobytes()
     )