[antonsoo]

Integer sample weights currently make weighted quantiles fail during CDF normalization:

import corner

print(corner.quantile([0., 1., 2.], [0.5], weights=[1, 1, 1]))
# Before: UFuncTypeError when dividing the integer CDF in place
# After: [1.0]

Accumulate the CDF with a dtype promoted to at least float64. This accepts integer lists/arrays and avoids integer accumulation overflow while preserving extended floating precision. The existing quantile convention is unchanged.

Regression tests cover integer lists, signed arrays and unsigned arrays. The complete suite passes on Python 3.12 and 3.14, with the existing skip, and all pre-commit hooks pass. On archived eight-schools posterior draws, integer-weight plots, titles and percentile ranges match the corresponding floating-weight results, including rendered images.