fix(parse): add parentheses for operator precedence in minItems check

[xodn348]

What

Adds explicit parentheses to the minItems condition in transform_schema() to clarify operator precedence, and adds missing test coverage for minItems: 0 and minItems: 1.

Why

The condition on line 149 of _transform.py:

if min_items is not None and min_items == 0 or min_items == 1:

evaluates as (min_items is not None and min_items == 0) or (min_items == 1) due to Python's operator precedence (and binds tighter than or). The intended grouping is min_items is not None and (min_items == 0 or min_items == 1).

Both produce identical results for all practical inputs (None, 0, 1, 2+), so this is not a behavioral bug — but the missing parentheses make the intent ambiguous and could mislead future readers or refactors.

Changes

• src/anthropic/lib/_parse/_transform.py: Add parentheses around the or clause
• tests/lib/_parse/test_transform.py: Add test_array_schema_min_items_zero and test_array_schema_min_items_one to cover the previously untested code path

Verification

• Build: pass
• Tests: pass (15 passed)
• Lint: pass
• Type check: pass (pyright 0 errors)

[kiwigitops]

This looks behavior-preserving rather than a parser fix. The original condition already only keeps minItems when the value is 0 or 1; min_items == 1 is false for None, so the added tests would pass on the current code too.

If the goal is readability, it may be clearer to frame this as a small cleanup. If there is an actual failing schema case, the regression test should cover a value that fails before this change.