fix(bigquery): Prefer query over table in get_table_query_string

[Jwrede]

Summary

Fixes #6200

When both table and query are set on a BigQuerySource, get_table_query_string() silently ignores query and always returns the table reference. This makes it impossible to use a custom read query (e.g., for deduplication via QUALIFY) on a PushSource, since PushSource requires table for offline writes.

Root cause: get_table_query_string() checks if self.table first — since table is always truthy when set, query is never reached.

Fix: Invert the priority — prefer query when present (it's more specific and intentionally provided), fall back to table. The write path (offline_write_batch()) accesses .table directly and is unaffected.

Also applies the same fix to get_table_column_names_and_types() so schema inference uses the query when both are set, matching the actual read path.

Changes

• bigquery_source.py: Swap condition order in get_table_query_string() and get_table_column_names_and_types() to prefer query over table
• test_bigquery.py: Add 4 unit tests covering table-only, query-only, both-set, and write-path-unaffected scenarios

Test plan

• All 11 existing + new unit tests pass (pytest sdk/python/tests/unit/infra/offline_stores/test_bigquery.py)
• Integration test with BigQuery PushSource (requires GCP credentials)

[Jwrede]

The failing test (test_online_write_batch_async_skip_dedup_single_pipeline in test_redis.py) is unrelated to this PR — it's an event-loop issue introduced in 2e50da0 that affects macOS and Python 3.12. My changes only touch bigquery_source.py and test_bigquery.py.

The 3.10 and 3.11 Ubuntu runs pass cleanly.

[ntkathole]

@Jwrede init from bigquery_source.py says Exactly one of 'table' and 'query' must be specified. ?

[Jwrede]

Good catch, the docstring says "Exactly one of 'table' and 'query' must be specified" but the actual validation (line 67) only enforces at least one:

if table is None and query is None:
    raise ValueError('No "table" or "query" argument provided.')

It has never rejected both being reads. I'll push an update to fix the docstring to match reality.