MetroLux exposures can be made very long by changing the counting mode to 10x. This allows for exposures of 9999 seconds or 167 minutes. This can be extended by 2x by using percent at 200%. This can be extended further (usually) by adjusting the calibration number on the photo sensors or by applying ND to the photo sensor. This gets you over 5 hours without any difficulty.
The MetroLux photo diode (and probably all diodes for this application) has a minimum response of 320 nm and a max of 730 nm. This max is way outside the response of any photo paper, silver or plat or other. However, it may be workable without filtration if the light (visible) is directly correlated to the UV. It probably is over a short period, but not day to day, as suspended particles in the atmosphere would greatly affect the ratio of visible to UV.
Adding a blue filter would help this correlation for two reasons. First and most is that the silicon photodiode is exceptionally responsive to red light, which you don't want. Second, by using blue, it eliminates the red leaving the blue and/or UV which are more closely correlated and have similar response.
Should the filter need to pass only UV, then the problem is more difficult for me. No doubt some combination of Rosco gels could get you a strong UV pass, but these are not spec'd much below 340 nm. I don't know how short of wavelength these 'other' processes 'see'. However, glass blocks the very short UV unless it is special or coated.
Should shorter wavelengths be required by the photo sensor, photo diodes that are enhanced are available to see to 190 nm. These are a little more expensive, but not prohibitive.
My guess is that nothing very fancy is required. In my dealings with alt process folks, standard silicon diodes have always been sufficiently repsonsive. The good news, with the sun, you have plenty of energy, plenty of photons.
