Please correct me if I'm wrong, but the mining difficulty is retargeted so that on average a block is found every 10 minutes. So it is regardless of the network hashrate.
Yes, but that difficulty is only adjusted ever 2016 blocks. So when 2016 blocks should take 14 days, in reality they might take, say, 12 days.
If the hashrate goes up even by 1000 times, the difficulty increases so that 1 block will be found in 10 minutes.
If it were to go up by a factor of 1000 in a single event, you're right, that would have very little impact- e.g. it would go from 14 days to 0.0014 days or about 20 minutes, and after that it would take 14 days again.
In reality, it goes up (or down) by a few percent, and does that for every block, and thus every difficulty adjustment and every reward halving.
Of course, until the difficulty increase happens you may be finding blocks sooner or later, but on long time average it should be 10 minutes.
'Should' being the key word there
I recently made a correction to that table and was considering removing the projected years - seeing it cited once again makes me think I really should do so
You made a correction? Where can I find it?
There was a single satoshi too many in some of the values - probably as a result of a floating point error in the original editor's output. You can find some further discussion about that particular topic in the Talk page.