feat: add ADC authentication

[yuriolive]

This pull request improves the authentication flexibility and metadata discovery reliability of tap-bigquery. The primary focus is aligning credential resolution with standard Google Cloud patterns (Application Default Credentials) and fixing issues where table names were being incorrectly prefixed during catalog discovery.

---

Key Changes

1. Robust Credential Resolution (ADC Support)

Introduced a centralized _get_bigquery_client utility and updated BigQueryConnector.create_engine to follow a standard resolution order. This allows the tap to work seamlessly in environments like GKE, Cloud Run, or Airflow using Workload Identity without requiring explicit configuration.

Resolution Order:

1. Config: google_application_credentials (can be a JSON string, a Python dict, or a file path).
2. Environment Variable: GOOGLE_APPLICATION_CREDENTIALS_STRING (JSON string).
3. Environment Variable: GOOGLE_APPLICATION_CREDENTIALS (Path to file).
4. Implicit ADC: Fallback to environment-based Application Default Credentials.

2. Improved Table Discovery & Normalization

• Per-table Reflection: Replaced bulk reflection with a one-by-one table discovery approach in discover_catalog_entries. This prevents the SQLAlchemy dialect from incorrectly resolving dataset names as project IDs during bulk operations.
• Name Normalization: Added _normalize_table_name to consistently strip dataset/schema prefixes from table names. This ensures that tap_stream_id and table identifiers remain clean and consistent.
• Default Selection: Discovered streams are now explicitly marked as selected-by-default in the metadata to improve the "out-of-the-box" experience for users.

3. Schema & Configuration Updates

• Optional Credentials: The google_application_credentials config property is no longer marked as required=True, as the tap can now rely on environment variables or ADC.
• Documentation: Updated README.md to clearly outline the new authentication flow.

---

Technical Details

• tap_bigquery/client.py: Refactored the client creation logic into a standalone function used by the stream's cached property. Added support for passing project_id to the BigQuery client to ensure correct billing and scoping.
• tap_bigquery/connector.py:
  • Updated create_engine to support credentials_path for file-based auth via SQLAlchemy.
  • Implemented discover_catalog_entries and _ensure_selected_by_default.
• tap_bigquery/tap.py: Updated JSON schema for google_application_credentials to reflect its optional status and added a more descriptive explanation for users.

---

How to Test

1. ADC Test: Run the tap in an environment with gcloud auth application-default login active and no google_application_credentials in the config.
2. String Config Test: Provide the service account JSON directly as a string in the config.json.
3. Discovery Test: Run tap-bigquery --discover and verify that tap_stream_id values do not contain redundant dataset prefixes (e.g., my_dataset.my_table should be my_table if the stream ID is already prefixed by schema).