BigQuery: Use configurable bucket name for GCS samples data in systems tests.

[tswast]

This allows the VPC-SC team to use their private mirror which is within
the security boundary when testing BigQuery VPC-SC support.

Closes #8550.

[tseaver]

Do we need to update noxfile.py to set this value in the environment?

[tswast]

No, the default of cloud-samples-data is fine for our purposes. The VPC-SC team has a separate Kokoro (or maybe something similar?) config where they'll set the necessary environment variable.

[tswast]

___________________ test_bq_to_arrow_array_w_special_floats ____________________
module_under_test = <module 'google.cloud.bigquery._pandas_helpers' from '/tmpfs/src/github/google-cloud-python/bigquery/google/cloud/bigquery/_pandas_helpers.py'>
    @pytest.mark.skipif(pandas is None, reason="Requires `pandas`")
    @pytest.mark.skipif(pyarrow is None, reason="Requires `pyarrow`")
    def test_bq_to_arrow_array_w_special_floats(module_under_test):
        bq_field = schema.SchemaField("field_name", "FLOAT64")
        rows = [float("-inf"), float("nan"), float("inf"), None]
        series = pandas.Series(rows, dtype="object")
        arrow_array = module_under_test.bq_to_arrow_array(series, bq_field)
        roundtrip = arrow_array.to_pylist()
        assert len(rows) == len(roundtrip)
        assert roundtrip[0] == float("-inf")
>       assert roundtrip[1] != roundtrip[1]  # NaN doesn't equal itself.
E       assert None != None

failure is unrelated. I think it's a behavior change / regression from pyarrow.

[tseaver]

@tswast How should we track that regression? All BigQuery CI is b0rked ATM.

[tswast]

@tswast How should we track that regression? All BigQuery CI is b0rked ATM.

Let's pin to pyarrow==0.13.0 in the noxfile until I can reproduce the regression and file an issue on the Apache Arrow JIRA.

[tswast]

Sent #8784.