Caller cancellation must stop recovery fallbacks

When request-level cancellation is passed into parallel primary work, aggregation must preserve it as a terminal condition rather than collapse it into generic technical failure that activates recovery. Otherwise a disconnected client can trigger another provider call after cancellation, wasting cost and producing misleading fallback telemetry. Propagate or rethrow caller abort separately from timeout and provider failure, check the signal before recovery, and pass it into every fallback transport. Provider work already accepted upstream may remain billable, so this limits new work rather than guaranteeing cancellation.

Applies when: parallel inference or service calls fall back to another provider or implementation. Keywords: AbortSignal cancellation fallback provider parallel request telemetry cost