So uhh, when are you gunna use a dataframe and when will you stick to SQL? My understanding is that SQL can sorta do a very powerful (and standardized/fast) subset of what a dataframe can