fix: update langfuse example to use correct TraceEmitter protocol

- Replace ContextTraceEmitter with TraceEmitter for proper ActionEvent protocol usage
- Add try/finally block for guaranteed sink cleanup and resource management
- Ensures example follows correct protocol contract and demonstrates safe resource handling

Co-authored-by: MervinPraison <454862+MervinPraison@users.noreply.github.com>

Author: praisonai-triage-agent[bot]

examples/observability/langfuse_example.py

@@ -13,14 +13,14 @@
 """
 from praisonaiagents import Agent
 from praisonai.observability import LangfuseSink
-from praisonaiagents.trace.context_events import (
-    ContextTraceEmitter, set_context_emitter
+from praisonaiagents.trace.protocol import (
+    TraceEmitter, set_default_emitter
 )
 
 # Initialize Langfuse observability
 sink = LangfuseSink()
-emitter = ContextTraceEmitter(sink=sink, enabled=True)
-set_context_emitter(emitter)
+emitter = TraceEmitter(sink=sink, enabled=True)
+set_default_emitter(emitter)
 
 # Create and run agent — all traces automatically captured
 agent = Agent(
@@ -29,11 +29,12 @@
     llm="openai/gpt-4o-mini",
 )
 
-result = agent.start("Write a Python function to calculate fibonacci numbers")
-print(result)
-
-# Flush traces
-sink.flush()
-sink.close()
+try:
+    result = agent.start("Write a Python function to calculate fibonacci numbers")
+    print(result)
+finally:
+    # Ensure traces are flushed and resources cleaned up
+    sink.flush()
+    sink.close()
 
 print("\nCheck Langfuse dashboard for traces!")