HITL forward _aura_reasoning to UX #277

We capture intent on all tool call in orchestration where the LLM is forced to explain why its calling a certain tool. Take this output and foward it onto our webhook schema in two spots:

1. HITL called as tool (agent mediation) - add _aura_reasoning and pass it through so the webhook receiver can understand WHY the agent is asking for approval
2. Auto tools flagged with a hard gate have _aura_reasonning from the original tool call that was intercepted with HITL- we can forward that for the WHY in the webhook schema