[Bug] Fix decode stats error on output_len 1

[HaiShaw]

Motivation

Fix error: Decode. median latency: nan s, median throughput: nan token/s from output_len=1, as first output token is a result of Prefill.

Modifications

• Added condition to Decode stats gathering.
• Increased warmup steps (output_len) minor.

Checklist

• [+] Format your code according to the Contributor Guide.
• [+] Add unit tests as outlined in the Contributor Guide.
• [+] Update documentation as needed, including docstrings or example tutorials.

[glen-amd]

Nit:

From the PoV of semantics, the check if output_len > 1 should go ahead of the line decode_latencies = [] because the decoding phase won't happen unless the specified arg (one of the values of BenchArgs.output_len) is greater than 1.