isSubordinateResult() intercepts TASK_FAILED before processSubordinateFailure()

[mvillmow]

ModuleLeadAgent::isSubordinateResult() returns true for any message with command == "response". TASK_FAILED messages also carry command == "response", so if the TASK_FAILED check in LeadAgentBase::processMessage() were ever reordered below the isSubordinateResult() check, failures would be silently treated as successes. The two checks are currently ordered correctly, but the coupling is fragile. Consider using command != "response" OR action_type == TASK_FAILED as the dispatch condition, or give TASK_FAILED messages a distinct command string (e.g. "failed").

Follow-up from #87

[mvillmow]

Already addressed by commit 7638b66 (feat(agents): handle TASK_FAILED events to prevent DAG deadlocks), merged to main.

Verified: include/agents/lead_agent_base_impl.hpp:47-48 — TASK_FAILED is now checked BEFORE isSubordinateResult(), so it is no longer intercepted by the wrong branch:

if (msg.action_type == core::ActionType::TASK_FAILED) {
    processSubordinateFailure(msg);  // line 47-48
    co_return ...;
}
if (isSubordinateResult(msg)) { ... }  // line 53 — never reached for TASK_FAILED

Closing as already-done.