Add failure reason tracking to compute actions and Ctrl+C scroll region cleanup

- Add FailureReason field to RunnerAction, propagated through /jobs/completed
  response so remote schedulers can see why actions failed
- Log failure reasons at point of failure: exit code, output gathering
  exceptions, sandbox setup errors, remote runner shutdown
- Log warning when retry limit is exhausted for an action
- Install Ctrl+C/Ctrl+Break handler to reset terminal scroll region when
  ProgressBar is active, preventing broken terminal state on interruption
- Clear FailureReason on action reschedule so retries start fresh

1 files changed, 1 insertions, 0 deletions

diff --git a/src/zencompute/runners/functionrunner.cpp b/src/zencompute/runners/functionrunner.cpp
index 64ff7fd50..f3d5acca5 100644
--- a/src/zencompute/runners/functionrunner.cpp
+++ b/src/zencompute/runners/functionrunner.cpp
@@ -400,6 +400,7 @@ RunnerAction::ResetActionStateToPending()
 
 	// Clear execution fields
 	ExecutionLocation.clear();
+	FailureReason.clear();
 	CpuUsagePercent.store(-1.0f, std::memory_order_relaxed);
 	CpuSeconds.store(0.0f, std::memory_order_relaxed);