Hi Tom
I dont have a Comet clutch handy, but suspect having another to measure up / compare is the easiest way forward.
There are a number of possibles that come to mind:
Why has only one clutch plate spun? Is that particular plate smaller(diameter) than the rest? If these are replacements for the 'original' plates, can you compare these? Has the drum been over-packed (too many plates)?
I'm used to the clutch on a slightly different Burman B52 box. On these, the plates are actually quite loose in the drum, that is sort of the point, allowing free movement.
Heavily used drums may have evidence of the edges of the clutch plate slots (dogs) having been filed smooth to remove notches / roughness to allow the plates to move forward and back freely. Quite a bit of this would be acceptable.
I do have more than 50% of the 'ears'(dogs) on my plates overlapping with the drum slots; but that wouldnt seem to be too much of a problem. The dogs aren't really 'deep' but somehow yours has got to a point with no overlap!
It seems unlikely that the inside of the drum would have been skimmed or machined in any way to improve things but its easily checked. There were actually 'earlier' and 'later' drums for Burman clutches, the plates are not interchangeable. I gather the slots were machined in the early drums, the later ones have short bars welded on to create the dogs. There were occasional problems with the early type of drum cracking at the back of the machined slots; this allows the sides of the drum to splay outwards - which could allow a plate to move?
Is the drum correct for the bike? There were Burman drums of various depth's to accommodate different numbers of plates.
Is the clutch centre actually dead centre within the drum? The Comet centre should have 5 studs, but these aren't always easy to find and 4 stud centres from different clutches (the 350 AJS/Matchless singles?) were sometimes adapted by having an extra hole drilled.
The inner face of the friction plates does not engage into the dogs on the clutch centre. This face should be free to rotate over the outer surface of the clutch ccntre; however, if there is an excessive gap then potentially a plate could be sitting so far off centre that the plates outer dogs could disengage?
Has the plate actually spun or just moved a bit / come out of its dogs?
Having just re-read my reply I'm not sure if I've helped or just asked a load of questions in reply to your question!
I'm intrigued, best of luck!