I tried to post this earlier and it appeats to have vanished. Hope it doesn't get posted twice.
Ok I missed the fact that the rockets weren't attached to the shell. Saw rocket assist and thought modern version of it.
However a couple of points. Hot gasses in general are corrosive. Even some older propelents such as black powder can be chemiclally as well as thermally corrosive. How much of a problem it would be in these guns I don't know but it could be a factor.
I didn't see anything in the two referances you gave that would indicate how accurate the big guns would be vs a terrestial target.
If you look at the info from the US 16" gun at:
http://www.navweaps.com/Weapons/WNUS_16-50_mk7.htm
it states
For example, during test shoots off Crete in 1987, fifteen shells were fired from 34,000 yards (31,900 m), five from the right gun of each turret. The pattern size was 220 yards (200 m), 0.64% of the total range. 14 out of the 15 landed within 250 yards (230 m) of the center of the pattern and 8 were within 150 yards (140 m). Shell-to-shell dispersion was 123 yards (112 m), 0.36% of total range.
Now given that it looks like the CEP for that shoot was around 150 yards at 17 nm, I'm guessing the lighter higher projectile mentioned here would be lucky to have a CEP less than 1 nm when fired on London.