Add TRANSACTION_COLUMNS and REFERENCE_COLUMNS config parameters

[kevinschaich]

Fixes #6

Transaction Columns

TRANSACTION_COLUMNS specifies a list of Plaid transaction properties (using _.get() syntax) to override the default automated columns. Each time you run Mintable, all the contents of these columns will be cleared and overwritten.

For example, if you only want to auto-populate the name and amount for each transaction, you could add the following line to your .env file:

TRANSACTION_COLUMNS=["name", "amount"]

Warning: Your mileage may vary if you choose to use additional properties outside the tested defaults (date, amount, name, account, category.0, category.1, pending). Proceed at your own risk, you're in uncharted territory.

Reference Columns

REFERENCE_COLUMNS specifies a list of additional, non-automated columns for your reference/bookkeeping purposes. Each time you run Mintable, the contents of these columns will be preserved.

For example, if you want to add one column to track work expenses, and another to track joint expenses shared with a partner, you could add the following line to your .env file:

REFERENCE_COLUMNS=["work", "joint"]

Warning: Since reference columns are not automated by Mintable, they have the potential to get out of sync with transaction data (for example, if your bank deletes a transaction, causing a row to get removed in TRANSACTION_COLUMNS)