Update `test_open_data_assets.py` script to handle historical mismatches

[jeancochrane]

The test-open-data-assets workflow fails basically every time due to a quirk in our data upload schedule: We only update historical (prior year) values once a year, and we only update the most recent year each month, but we test all years every month. This means that historical years will continue to fail throughout the year as our staff make small edits to historical years (example).

Two options we have for handling this:

1. Change historical tests to a percentage threshold (e.g. within 5% different instead of exactly matching)
2. Only test historical values once per year, when we update them

Or maybe a combination of these two approaches.