I don't see too many issues using a standard USB camera in a vacuum. Your two largest issues are likely to be heat dissipation, and outgassing from the camera components.
Depending on how high a vacuum you need, you may need to package the camera in a hermetic container. There is almost inevitably going to be at least a small amount of soldering flux/VOCs left on the PCBs from production, and it will outgas when you pump the system down.
If you have sufficient pump capacity, or don't need extremely high vacuum, it shouldn't be too much trouble.
You could also try removing all the plastic from the camera, and baking the PCB assembly at a low temperature to speed up the out-gassing (~100°C?).
Second, at least some of the USB-cameras I have used do dissipate a fair amount of power, so depending on the camera, you may have localized heating issues due to the lack of convection cooling. It should be possible to deal with this using proper heat-strapping, though I think it would probably better to just pick a device with low dissipation.
Lastly, you can certainly buy some pretty interesting USB cameras, which do support all sorts of fun things like variable ISO, and variable shutter time. It mostly depends on your budget.
If you don't mind experimenting, I would say go to your local electronics store, and just buy a webcam, and see what happens (and post pictures!).
The hardest part, I think, is probably going to be hermetically sealing the cable feed-through. You can't just stick the cable in a cable gland, as you will get leakage through the cable, both between the strands, and between the individual wires and the sheath. You will need hermetically rated bulkhead connectors.
There are various methods to determine the the number of electrons collected in a given sampling interval (look up - photon transfer curve) but this in itself does not give you what the QE (Quantum Efficiency) is of the device. QE is the measure of how many electrons are generated per photon incidence (always less than 1.0 and in CCD's typically in the 28% range).
You must note that the QE also varies with wavelength as the various dielectric layer thicknesses above the photodiodes emphasize and attenuate at different wavelengths.
The only real answer is that you must calibrate the sensor against a known source or a calibrated detector. There are NIST calibrated detectors available in various supply companies. These devices will have a calibration curve vs. wavelength that you can use to calibrate your device.
Here is a link to calibrated PD's on Thorlabs - search on "Calibrated Detectors". They have detectors that have built in amplifiers and bare detectors even with fibre connections. There are many others that sell them too.
Best Answer
CCDs have several types of noise, but only two are relevant here:
Read noise. This is the integration time independent component of the read out process. Every time you read out the circuit you get this noise.
Dark current. These are randomly generated thermal electrons that contribute additional (non-photon) shot noise. As such they are proportional to integration time.
From these it should be obvious that multiple read outs are going to have more read noise and thus will not be better. Less obviously they may not be worse if your dark shot noise component swamps the read noise, but this is uncommon unless your integration time is very long.
For example, a contemporary midrange sensor (which will probably be CMOS since almost no one makes CCDs anymore), read noise might be 4 electrons and dark current 10 electrons per second. The dark shot noise for 2 seconds would be sqrt(20) electrons. Thus if you read out a few shorter exposures you would quickly swamp the dark noise with read noise.
Note that averaging multiple read outs does increase dynamic range, so if you have a very bright signal this may still be a good idea.