/**
* Resolve provider-aware automatic reasoning options for an agent loop.
*
* Harn keeps the provider calibration in the VM so raw `llm_call`,
* `agent_loop`, and ACP session config all share one policy vocabulary:
* `auto`, `off`, `minimal`, `low`, `medium`, `high`, and `xhigh`.
*
* @effects: []
* @allocation: heap
* @errors: []
* @api_stability: experimental
* @example: agent_reasoning_apply(opts)
*/
pub fn agent_reasoning_apply(opts) {
return llm_apply_reasoning_policy(opts)
}