feat: add pandas-gbq to optional extras

[shuoweil]

Adds pandas-gbq>=0.35.1 and pyarrow>=3.0.0 to the pandas optional extra in google-cloud-bigquery-storage, while keeping pyarrow extra lightweight without pandas dependencies.

Details

• Updates extras["pandas"] in setup.py to include pyarrow>=3.0.0 and pandas-gbq>=0.35.1.
• Keeps extras["pyarrow"] as pyarrow>=3.0.0 only so pure Arrow/Polars/DuckDB workflows avoid pulling in pandas and numpy.

Fixes #<526614511> 🦕

[gemini-code-assist]

Code Review

This pull request updates the setup.py file for google-cloud-bigquery-storage to add pandas-gbq and pyarrow to the pandas extra, and pandas-gbq to the pyarrow extra. The reviewer points out that adding pandas-gbq is unnecessary and introduces a circular dependency loop, as pandas-gbq depends on google-cloud-bigquery, which depends on google-cloud-bigquery-storage.

[gemini-code-assist]

Adding pandas-gbq as an optional extra dependency for pandas and pyarrow in google-cloud-bigquery-storage is unnecessary and introduces a circular dependency chain.

1. No Direct Usage: google-cloud-bigquery-storage does not import or use pandas-gbq directly. It only requires pandas and pyarrow (or fastavro) to serialize/deserialize data streams.
2. Circular Dependency: pandas-gbq depends on google-cloud-bigquery, which in turn depends on google-cloud-bigquery-storage. Adding pandas-gbq here creates a circular dependency loop that can cause issues during dependency resolution with package managers like pip or poetry.

    "pandas": [
        "pandas>=1.1.3",
        "pyarrow>=3.0.0",
    ],
    "fastavro": ["fastavro>=1.1.0"],
    "pyarrow": [
        "pyarrow>=3.0.0",
    ],