1. Given outliers in gains (30% - 100% ETH % Gain over HODL) I checked each of their LPs and fees manually to verify that these positions had the liquidity removed w/o being withdrawn. This inflates fees and b/c fee_correction() only works on same-transaction hash (a safety feature) I manually aligned the liquidity remove <=> liquidity withdrawn <=> Fee collected transactions.
Fee subtraction harder to do when transactions include multiple add,remove,collect,remove,add,collect all in the same transaction-pool-nft_id or tx-pool-ticklower-tickupper-nf_position_manager
Will correct here outside of asking data engineering to fix as there are a lot of nuances with event indexes that could mess things up.
Alerted DE team; really rare, effectively just need to add up all fees so that the fee table has 1 row per transaction_hash-POOL observation. Then remove double count once.
