I test with the Eneloops right off the charger (within 20 minutes usually). I would think that the voltage drop when the light is turned on will be much larger than letting them rest for a couple of hours. That’s why I included the 60s number. The light temperature will have a bigger effect than the battery voltage, since the output drops very rapidly at the beginning but not at all after the temperature has settled even though voltage keeps dropping.
It seems your readings are very stable between different tests.
Meters react very differently to different wavelenghts as djozz’s latest lux meter test shows. For example, the noname LX-1330B reads 7.5% low on the 4500K 70 CRI Supwildfire Mitko thrower and 3% high on the 4000K 90+ CRI Sofirn SP33 compared to a good meter.