Cost analysis and improvement to ethereum harvesters

Currently we are harvesting based on the base gas price and time since last harvest. We should investigate using the estimated outstanding harvestable amount based on the base APR, value in the vault, and time since last harvest to determine whether a harvest makes economic sense right now. Some of the current setts have such little TVL and base APR that harvests don't make sense at 3 day intervals (and might not even make sense at 5 or 7 day intervals, more research required).

Google docs: https://docs.google.com/document/d/1tJTrxUFXrE8cr4dTOvjb9KO8reXPlrnFXP5EHW0utCc/edit?usp=sharing