ArrayUtils min and max - initialValue

For initial values when doing max or min, we sometimes use user-provided initial values, sometimes explicitly assume a double type (like -DBL_MAX), and sometimes use std::numeric_limits host only, standard library not available on the device). We should make this all consistent. Perhaps we could make something like std::numeric_limits that works on the GPU?

Start by locating the ArrayUtils min and max implementations and their host/device call sites. Compare the user-provided initial values, -DBL_MAX, and std::numeric_limits usage, including paths where the standard library is unavailable on the device. Done means selecting and applying one consistent initial-value approach that works across host and GPU code, with relevant tests updated or added.

Written by the indexing model from the issue text.