In Proof-of-Work, a block hash is costly to manipulate (say, wanting the hash to be even, or have the hash of the hash smaller than some number) since there is a lot of computing power put into selecting the right header to produce a valid hash.
I'm wondering - are the block hashes in the various Proof-of-Stake implementations easier to manipulate since the hashes aren't bound by PoW mining? Could a PoS miner spend some computational time to generate many valid blocks and pick one to use, or are there some other constrains preventing block hash fudging?