I said I would report back, so here goes...

To cut a long story short, the problem has gone away; the long story is as follows. Note that all the scanned prints given as examples below are scanned against a sheet of white paper as a reference:

(a) Initially, prints looked like scan0001.jpg. This is a border-less test print, held in place by hand. It was as though the natural colour of the Fuji Crystal Archive paper had not been removed.

(b) Over the course of time (a few days, maybe, with the same chemicals remaining in the Nova slot processor), the strong cyan colour disappeared on subsequent prints so that prints then looked like scan0002.jpg. This is also a border-less test print, held in place by hand. Note that this is without changing chemicals or filtration from scan0001.jpg. The prints were developed in a vertical slot processor and the bottom edge of the prints is to the left, where there is still a cyan line, as though chemicals have been caught on the edge of the paper rather than draining off.

(c) After a few more days, the cyan problem largely disappeared but there remained an overall greeny-yellow cast as seen in scan0003.jpg. The bottom edge is to the left and the colour cast increases in intensity towards the left.

(d) At this point, I made the first post in this thread. After considering all the comments, I decided to insulate the Nova tank (the developer was in an outer slot and so slightly less warm than the stop bath and blix) and I decided to omit the stop bath altogether since it was citric acid based. I also replenished the developer and blix by draining off 250 ml of each and replacing with more of the original 5 litres (I'd done about 20 10x8 sheets, not 40 as stated previously). Each slot holds about 2 litres. The next time I began printing, I made a complete hash of getting the paper from the developer into the blix (no stop bath), so that in total it was in contact with developer for about 1 minute 30 seconds rather than the recommended 45 seconds at the temperature I was running at. The resulting print had mucky and stained borders and looked a mess. But this suggested that I might have been giving prints too much development and a 45 second development time (including draining which had been taking longer than I thought) produced spotless white borders with no stop bath. I have since replaced the stop bath with Kodak's acetic acid based one and now get good colours and white borders as in scan0004.jpg (the scan makes them look less white than they are).

So I'm inclined to conclude that the problem was some sort of chemical staining, but since I was able to get staining as described above without any stop bath it's not clear what additional effect the citric acid stop bath might have had. What I can say is that by following recommended developing times and temperatures as accurately as I can and with an acetic acid stop bath, I haven't (yet) had any more problems.

I'd still be interested to know why there was such a strong cyan stain initially. Did I not mix the chemicals well enough to start with?