Include token count in api.Chunk

And calculate the tokens/chunk for gemini responses, fixing the tok/s
meter for gemini models.

Further, only consider the first candidate of streamed gemini responses.

pkg/api/provider/openai/openai.go

@@ -319,8 +319,9 @@ func (c *OpenAIClient) CreateChatCompletionStream(
 			}
 		}
 		if len(delta.Content) > 0 {
-			output <- api.Chunk {
-				Content: delta.Content,
+			output <- api.Chunk{
+				Content:    delta.Content,
+				TokenCount: 1,
 			}
 			content.WriteString(delta.Content)
 		}