The TSL2561 also measures infra red. It has two diodes, one for visible and infra red light and one for infra red light only.
To my opinion, BH1750 fits most requirements, whereas the TSL2561 offers a wider spectrum, more sensitivity and more accuracy.
May I put the question itself in question: What do we want to measure? Is it really luminosity? To my understand those regular low-cost sensors report the illuminance (luminosity/brightness) in Lux.
From the meteorological point of view the irradiance (the transferred energy) would be more interesting, stated in Watts per area. This parameter is used to determine if there is “sunshine” (>120W/m² in upright position as of WMO). One can’t convert Lux in W/m².
But there is a cool DIY-approach! What might be a sensor for “getting the watts per area sent by the sun”? Of course the short-circuit current of a solar-panel! Brilliant:
I still don’t see an approach to compensate the “upright-position”-problem. May be a compensation taking the current position of the sun relative to sensor into account could be an approach. Or the usage of a big fish-eye lens. Or using multiple cells.
Good to have asked you @wtf! :-) My question behind is: What light conditions are good / bad for plants or for bees. So irradiation could be a more fitting value than luminance.
as always, a good start for reading (in german) about meteorological measurements is DWD’s guideline #6 on ground sensors, chapter 14 on solar-radiance, starting on page 141. Some renewal of the understanding of direct, diffuse and global radiation helped me a bit.
Last year a bachelor thesis was published at the university of Twente: Designing a low-cost autonomous pyranometer (PDF). The autor did some comparsion of low-cost sensors with a Davis instrument (spoiler: in the end it turned out this reference was a bit faulty), and decided a path using the combination of two low-cost sensors, Si1145 and ML8511.
Regarding sensors for measuring illuminance in lux – if that might be still of interest – I just discovered that Rob Tillart was updating the Arduino driver for the MAX44009 I2C lux sensor just recently.