Measuring Retrieval Robustness in LLMs, Training Small LLM Agents for End-to-End RAG, and More!