Bug fixes (#173)

This PR fixes two small bugs.
- Double counting of prompt tokens in the calculation of total
tokens_per_second
- The warmup percent was not working when using max duration (worked for
max requests). Traced this to a - which should have been a + in
benchmark/aggregator.py

Author: dagrayvid

src/guidellm/benchmark/aggregator.py

@@ -403,7 +403,7 @@ def add_result(
         in_warmup_duration = (
             self.args.warmup_duration
             and result.request_info.worker_start
-            <= (global_start_time - self.args.warmup_duration)
+            <= (global_start_time + self.args.warmup_duration)
         )
 
         if in_warmup_number or in_warmup_duration:

src/guidellm/benchmark/benchmark.py

@@ -816,7 +816,7 @@ def from_stats(
                         for req in total_with_output_first
                     ],
                     iter_counts=[
-                        req.prompt_tokens + req.output_tokens
+                        req.output_tokens
                         for req in total_with_output_first
                     ],
                     first_iter_counts=[