aromring If you read the Si1132 data sheet closely you would note that while the Si1132 is called a UV light sensor, it is actually a photodiode responsive to visible and infrared light, and calculates a "UV Index" value based on a modeled relationship between the photo diode's measurement measurement of light intensity and typical sunlight conditions at the earth's surface. What you will be measuring with it is light intensity across visible and infrared light (given its response curve on page 10 of the data sheet), not any specific UV frequency band. The sensor is far more interesting for measuring light intensity in general than knowing whether you are going to get a sunburn on orbit (which is what UV index is used to indicate on earth).
The Si1132 has 100 milli-lux resolution in it's light measurements, with a 1 lux to 128 kilo-lux dynamic range. Bright sunlight is 111 klx, and I would expect bright sunlight in space because, well, there are no clouds. If all the earth reflects is 3-5% of the sunlight as you quote in the worse case, that means the sensor would get 3 to 5 klx of light. The dynamic range of the sensor is still sufficient to capture that amount of light. One thing, though, you need to configure sensors for the expected lux range, as the photodiode's measurement can easily saturate if configure for low light conditions and the sensor is exposed to full sunlight. This is why in the flight software I am writing I have enabled the ability to change the sensor's configuration while on orbit through TTN downlinks.
The orientation of the satellite would, of course, could be a problem. But there in lies an interesting challenge: can you estimate the orientation of the satellite based on the combination of the magnetic field measurement (from the LSM9DS1), the orbital ephemeris, the time of the telemetry collection, and the light intensity measurement. If you are expecting LandSat type images from a $200 picosat, well, this may not be the right project for you. However, if you are the type to get into the gritty details of things, there is plenty of opportunity to get something interesting from this project.
It maybe true that the AmbaSat project may not be a "picosat for the masses" type project, as it is not really "plug and play". And it is true that not all sensors offered are fully useful in the space environment (for example, the BME680's humidity and pressure measurements are useless, but the temperature would still be useful). As illustrated, the Si1132 does have good potential for interesting results, and with some modeling and math, you should be able to extract interesting insights out of the measurements.
To me, the real issue for this project is the radio communications, specifically the frequency permits and how that will impact the hardware we need to fly. But having launched satellites professionally, I know that you don't have every detail solved at the start of the project to still have a successful flight. If it ultimately doesn't work out, its still going to be a fun journey.