Adding TPOT and ITL metrics

[Edwinhr716]

Added Time Per Output Token (TPOT) and Inter-Token Latency (ITL) metrics. In order to get these metrics, need to add the --stream-request flag.

Sample output:

====Result for Model: deepseek-ai/DeepSeek-R1====
Errors: {'ClientConnectorError': 0, 'TimeoutError': 0, 'ContentTypeError': 0, 'ClientOSError': 0, 'ServerDisconnectedError': 0, 'unknown_error': 0}
Total time: 115.56 s
Successful/total requests: 120/120
Requests/min: 62.30
Output_tokens/min: 13539.97
Input_tokens/min: 5607.26
Tokens/min: 19147.23
Average Time to First Token (s): 0.51
Average Inter-Token Latency (s): 0.08
Average Time Per Output Token (s): 0.08
Average seconds/token (includes waiting time on server): 0.06
Average milliseconds/request (includes waiting time on server): 18217.93
Average milliseconds/output_token (includes waiting time on server): 83.83
Average input length: 90.00
Average output length: 217.32