Added torchao nightly workflow (pytorch#128152)

Summary:
Add torchao benchmark workflow, upload the artifacts to GHA.

X-link: pytorch/benchmark#2273

Test Plan:
```
python run_benchmark.py torchao --ci
```

Differential Revision: D58140479

Pulled By: xuzhao9

Pull Request resolved: pytorch#128152
Approved by: https://github.com/jerryzh168

benchmarks/dynamo/common.py

@@ -3975,9 +3975,12 @@ def run(runner, args, original_dir=None):
             assert "cuda" in args.devices, "Quantization requires CUDA device."
             assert args.bfloat16, "Quantization requires dtype bfloat16."
             try:
-                from .torchao_backend import setup_baseline, torchao_optimize_ctx
-            except ImportError:
                 from torchao_backend import setup_baseline, torchao_optimize_ctx
+            except ImportError:
+                from userbenchmark.dynamo.dynamobench.torchao_backend import (
+                    setup_baseline,
+                    torchao_optimize_ctx,
+                )
 
             setup_baseline()
             baseline_ctx = functools.partial(