I think this has been covered, but unfiltered is definitely preferable to a UVIR blocking filter.
For my part I tend to use unfiltered for time-series work (exoplanets previously, asteroids these days) to achieve the greatest SNR and the fastest cadence. As only the minute to minute variation is of interest there is no great need to have the observed magnitudes be comparable with measurements made by other observers.
When making precision measurements I almost always use a V filter but have been known to submit CV measurements — unfiltered with comparison magnitudes given in Johnson V.
No great need for a filter, in other words, except for red stars as noted above. A filter is nice to have, undoubtedly, but no-one should let expense put them off making scientifically valuable measurements.
One day I really must see if I can find a round tuit down the back of the sofa because there are literally thousands of images on disk which were taken with the Seestar’s OSC camera. Trouble is that all of them need de-Bayering and the green channel measurements transformed to standard V. Another example of where external filters are no needed — the Bayer mask consists of filters which are internal to the detector.