perf: eliminate redundant getattr calls in token metrics collection

- Remove duplicate llm variable definitions in aexecute_task and execute_task
- Reuse existing llm variable instead of redefining with getattr calls
- Reduces function call overhead without changing functionality
- Addresses code duplication identified in PR review

Co-authored-by: Mervin Praison <MervinPraison@users.noreply.github.com>

Author: github-actions[bot]

src/praisonai-agents/praisonaiagents/agents/agents.py

@@ -414,7 +414,6 @@ def _get_multimodal_message(text_prompt, images):
             )
 
             # Add token metrics if available
-            llm = getattr(executor_agent, 'llm', None) or getattr(executor_agent, 'llm_instance', None)
             if llm and hasattr(llm, 'last_token_metrics'):
                 token_metrics = llm.last_token_metrics
                 if token_metrics:
@@ -774,7 +773,6 @@ def _get_multimodal_message(text_prompt, images):
             )
 
             # Add token metrics if available
-            llm = getattr(executor_agent, 'llm', None) or getattr(executor_agent, 'llm_instance', None)
             if llm and hasattr(llm, 'last_token_metrics'):
                 token_metrics = llm.last_token_metrics
                 if token_metrics: