So a sudden 40% loss in hashrate could substantially impact & lengthen the 10min/block rate, unless there's quite a bit of excess capacity (equivalent to 40%) that could be brought online in response.
You have to spend 40% more time before someone finds a valid hash, but any miner is 40% more likely to be the one to find a valid hash... sounds like that should cancel out indeed but I'm not entirely sure.
You (as an individual miner) would still mine the same number of blocks on average as before, nothing has changed from your perspective (difficulty/hashrate). You would find a larger percent of all blocks mined, but same absolute number of blocks.
Does anyone know if there are good reasons for not adjusting the difficulty more often?
Yes, it could allow to cheat the algorithm. Consider the two weeks period as a strong low-pass filter on cheating tentative.