API: Add timings to usage stats

It's useful for the client to know what the T/s and total time for
generation are per-request.

Works with both completions and chat completions.

Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>

endpoints/OAI/types/common.py

@@ -1,7 +1,7 @@
 """Common types for OAI."""
 
 from pydantic import BaseModel, Field
-from typing import Optional
+from typing import Optional, Union
 
 from common.sampling import BaseSamplerRequest, get_default_sampler_value
 
@@ -10,8 +10,13 @@ class UsageStats(BaseModel):
     """Represents usage stats."""
 
     prompt_tokens: int
+    prompt_time: Optional[float] = None
+    prompt_tokens_per_sec: Optional[Union[float, str]] = None
     completion_tokens: int
+    completion_time: Optional[float] = None
+    completion_tokens_per_sec: Optional[Union[float, str]] = None
     total_tokens: int
+    total_time: Optional[float] = None
 
 
 class CompletionResponseFormat(BaseModel):