Why might a RAG system with perfect Context Recall still fail to produce accurate responses? How does the F…

Why might a RAG system with perfect Context Recall still fail to produce accurate responses? How does the Faithfulness metric help diagnose this issue?