PR Description

This PR extends numpy.ptp (peak-to-peak: max - min) to support the where= keyword argument, consistent with other NumPy reduction functions such as sum, min, and max.

Changes Made

• Updated _methods._ptp to forward where into umr_maximum and umr_minimum.
• Extended public API in fromnumeric.py to accept where=np._NoValue.
• Updated docstrings to document the new where parameter.
• Updated type hints in fromnumeric.pyi to include where.
• Added unit tests in test_numeric.py (test_ptp_where) to validate functionality.

Behavior

• np.ptp now honors masks passed via where, enabling selective element inclusion in peak-to-peak computation.

• For example:

  >>> import numpy as np
  >>> a = np.array([1, 5, 7, 0])
  >>> where = [True, False, True, False]
  >>> np.ptp(a, where=where)
  6

• Edge cases (e.g., all-False masks) follow the identity rules of maximum (-inf) and minimum (+inf), resulting in -inf for floating-point arrays.

Testing

• Added direct test for masked computation.
• Verified broadcasting behavior and compatibility with existing parameters (axis, out, keepdims).

Documentation

• Added where parameter description to the ptp docstring.
• versionadded tag should be included to mark this feature.

This aligns ptp with other NumPy reductions and improves API consistency.

Fix #29675