Columns in schema rules are assertions about the requirement level of what headers should be present in a tsv file. Examples in specification: schema/rules/tabular_data/*

For each column in a rule.tabluar_data check we generate an error if the column is missing from the tsv and listed as required by the schema, a warning if the schema rule is clobbered by the sidecar but shouldn't be. If the column is not in the tsv we bail out and move to the next column, otherwise we type check each value in the column according to the type specified in the schema rule (or sidecar type information if applicable).