Add datasource_id wherever there's a sourceaccessioncode, and map to it (implicitly?)
This eliminates the need for materialized "breadcrumbs" in the staging tables, because the immediate parent can be looked up with just the input table's fkey and datasource name
Allow an input to appear multiple times in the same map spreadsheet (issue #463)
Best to do this before removing distinct tables, so that collisions requiring _alt are visually obvious
Remove distinct tables in map spreadsheets
Should come before Flatten VegX because once VegX is flat, there won't be a concept of mappings "tables" other than prefixes for column names
Flatten VegX to DwC-like standard column names
Use DwC, then VegX, then VegBank as the "standard" names
source of the term will be tracked in separate "Source" column in the main map spreadsheet
also incorporate datasource's custom names when these are unambiguous
"grab bag" of terms, like we currently do with DwC
Allow data provider to supply custom ordering of tables (hierarchical levels)
Flatten XML files to CSV for use with staging tables
Translate remaining XML functions to SQL functions
Filter correctable errors in errors tables
Parse date ranges
Date with unknown day -> middle of month: 00, XX, xx, ??
or record uncertainty? would require indeterminate dates in Postgres
Data provider feedback: Query to translate value-by-value table to per-record
just provide (e-mail) CSV to data provider; don't need web interface
Detect frameshift errors and discard those rows
DwC lat/long with " for seconds throws off CSV parsing
Need better ways to deal with non-standard CSV dialects
Heuristic field start/end detection to ignore embedded quotes?
