equal_interval crashes with ZeroDivisionError on constant input

[brendancol]

Problem

equal_interval() raises ZeroDivisionError when the input has max == min (all pixels equal, modulo NaN/inf). Other classifiers handle this case: std_mean collapses to a single class (see test_std_mean_all_same_values).

Reproduction

import numpy as np
import xarray as xr
from xrspatial.classify import equal_interval

agg = xr.DataArray(np.full((4, 5), 7.0))
equal_interval(agg)
# ZeroDivisionError: division by zero

Root Cause

_run_equal_interval in xrspatial/classify.py:890-892:

width = (max_data - min_data) / k
cuts = np.arange(min_data + width, max_data + width, width)

When max_data == min_data, width = 0 and np.arange(x, x, 0) raises ZeroDivisionError.

Fix

Short-circuit the degenerate case: if max_data == min_data (or min/max are non-finite), assign every finite pixel to a single class.

Scope

Affects numpy, cupy, dask-numpy, and dask-cupy backends (shared _run_equal_interval).

Found by

Security sweep (Cat 3 - NaN/Inf logic error, MEDIUM severity).