Copy source before modifying it in GroupBy

[earangol-stripe]

Summary

Copies the sources passed to a GB before applying modifications (normalizing and sanitizing).

There is special handling for join sources because the reference to the Join must be maintained to set the object name correctly from GC referrers.

Why / Goal

To prevent unexpected modifications to sources when shared across Joins/GBs.

Sources may be reused across Joins and GBs. The sanitize code here modifies the source in-place, which could cause unintended behaviors for GBs that reuse the source and don't expect those modifications. One specific example:

• group_by1 uses source1. One of its aggregations has input_col1 as the input column. If that column is not present in source1, _sanitize_columns will add it to the query selects.
• group_by2 also uses source1. It does not have input_col1 in any of its aggregations. If group_by1 was loaded before group_by2, its source would include that column unexpectedly.

Test Plan

• Added Unit Tests
• Covered by existing CI
• Integration tested

Checklist

• Documentation update

Reviewers

[pengyu-hou]

LGTM

Could you also add a unit test?