2. Can timestamp manipulation still be used profitably in low-hash forks?
Do you mean Bitcoin forks?
I believe it is not profitable for most of forks because most of Bitcoin forks die.
3. Would per block difficulty adjustment ever be worth considering?
Surely it does not worth considering by two reasons.
Network hashrate does not increase too much between two consecutive blocks.
Blockchain difficulty consequently does not jump up exponentially after one block.