sun_position: no sunrise/sunset or solar-noon helper for chronolocation workflows

[cognis-digital]

Summary

geolens.core.sun_position(lat, lon, when) gives the instantaneous azimuth/elevation/declination, and estimate_latitude_from_shadow assumes the observation is at (or near) local solar noon. But there is no way to find solar noon (or sunrise/sunset) for a location and date — so the analyst has to guess the noon instant, which is the single biggest error source in the shadow-stick method.

Motivation

Two very common OSINT/verification questions have no direct API today:

1. "When was local solar noon here on this date?" — needed to time-align a shadow-stick measurement (the method degrades quickly away from true noon).
2. "Was the sun even up when the photo claims it was taken?" — currently you have to sample sun_position across the day and look for the elevation sign change (see demos/08_sun_track_day.py, which does exactly this by hand).

Proposed addition (additive, std-lib only)

def solar_noon(lat, lon, date) -> datetime          # UTC instant of max elevation
def sunrise_sunset(lat, lon, date) -> dict           # {"sunrise": dt|None, "sunset": dt|None, "polar_day"/"polar_night": bool}

Both are derivable from the existing NOAA eq_time / hour-angle math already in sun_position — solar noon is 12:00 - eq_time/60 - lon/15 (UTC); sunrise/sunset from the standard hour-angle at the -0.833 deg horizon, with explicit None for polar day/night. No new dependency.

This would let estimate_latitude_from_shadow optionally snap to computed solar noon instead of trusting assume_local_noon, and give the verification demos a first-class daylight check.