b> The chances are, this is just caches being a bit slow to clear. A
b> normal refresh would probably have caused it all to appear as it
b> should.
Indeed, it turns out the problem "went away by itself", without even
before me needing to push the purge buttons offered to me. So it is
still a problem (race conditions), but one that doesn't last for long.
P.S., Odd that one needs to click to confirm purges: alas this also
thus requires a perl LWP script then to accomplish in batch mode.