BigQuery External Table Bad Rows Issue

I have a BigQuery table that's configured to be an External Table looking at Cloud Storage. The Source URI is: gs://[bucketname]/test_file*

I did not specify maxBadRecords in my table creation request, and the default value is 0. I also did not specify ignoreUnknownValues, and the default for that is False. So any extra, or fewer fields in my source json files would result in a "bad row", and an error should result. The schema for the table is:

[
  {
    "name": "second_column",
    "mode": "NULLABLE",
    "type": "INTEGER",
    "description": null,
    "fields": []
  },
  {
    "name": "first_column",
    "mode": "NULLABLE",
    "type": "STRING",
    "description": null,
    "fields": []
  }
]

I have 3 files:

test_file_1_1_1_1_1_1.json:

{"first_column": "a good value goes here", "second_column": 9}

test_file_2_2_2_2_2_2.json:

{"first_column": "a good value goes here", "second_column": 9, "extraneous_column":  "uh oh"}

test_file_3_3_3_3_3_3.json:

{"first_column": "a good value goes here"}

So file 1 matches the schema, file 2 has an extra column, and file 3 is missing a column.

Querying the table when only file 1 is in the bucket returns the expected data from the file. When I add file 2 to the bucket, I get this result:

which is not what I'd expect. When I also add the third file, I get:

Is this expected behavior? I'm confused, as I'd expect any number of bad rows to result in such an error.