update methods to estimate suppressed data

[catherinebirney]

• when estimating suppressed data, we now account for values of grandchildren
• new validation check to compare flow totals in a df before and after attribution
• updates to MLU functions to work for state methods
• update to substitute_nonexistent_values() to include all possible sectors at the state level
• drop 'AG LAND' mapping from USDA CoA Cropland Crosswalk because AG LAND (11) does not represent the summation of 111 and 112
• revisions to state water, state land, national land FBS yamls

[bl-young]

I successfully tested GHG_national_2017 and TRI_DMR_state_2012. The former shows changes (expected due to use of employment FBS) while the latter does not.

I did not review in depth your adjustments.