g1  correct=N  verified_flag=Y  attempts=1  
g2  correct=N  verified_flag=Y  attempts=2  
g3  correct=N  verified_flag=N  attempts=1  
g4  correct=N  verified_flag=N  attempts=1  
g5  correct=N  verified_flag=N  attempts=3  StructuredOutputError: No valid SQLDraft after 3 attempts
g6  correct=N  verified_flag=N  attempts=1  
g7  correct=N  verified_flag=N  attempts=1  
g8  correct=N  verified_flag=Y  attempts=1  

accuracy 0/8   wrong-but-verified: 3
