Owlgorithm: Supporting Self-Regulated Learning in Competitive Programming through LLM-Driven Reflection