Enriching single-arm clinical trials with external controls: possibilities and pitfalls