fix(agentchat): use running loop for sync input callbacks

[Ghraven]

Summary

• replace asyncio.get_event_loop() with asyncio.get_running_loop() when sync user-input callbacks are executed from async AgentChat paths
• update both UserProxyAgent._get_input() and UserInputManager so they use the already-running event loop before dispatching work to the executor

Problem

These code paths run inside coroutines and only need the active event loop to call run_in_executor(). asyncio.get_event_loop() is a legacy lookup API whose behavior has become stricter across Python versions, while get_running_loop() is the intended API when a coroutine is already executing.

Before / After

Before: sync input callbacks in AgentChat and Console used asyncio.get_event_loop() from async functions.

After: the same callbacks use asyncio.get_running_loop(), preserving behavior while avoiding the legacy event-loop lookup.

Verification

• python -m compileall python/packages/autogen-agentchat/src/autogen_agentchat/agents/_user_proxy_agent.py python/packages/autogen-agentchat/src/autogen_agentchat/ui/_console.py
• git diff --check
• uv run pytest tests/test_userproxy_agent.py -q (5 passed)