Why wouldn't it make sense to create a different server seed beforehand for each game and after it is used, immediately publish it?
Just wonder how will you check it? You decide to check our honesty - you check the client seed, remember it, you make a bet - and check what was the server seed - seems ok? But it isn't in this scenario we could just easily fool you by using any seed that will make any outcome we want - just enumerate few random seeds that would give us desired outcome. Even if we would somehow publish a hash of this seed - you cannot be sure that hash you see is for the seed that you will get when you bet, since any other player might use this seed before you.
But if a player uses that hash just before me, I could easily confirm this on the right side. Now you can say: doesn't prove anything since we could already do this.
Well with your current implementation you could also skip hashes from the batch and add fake bets to the right to compensate.
Sorry, but I don't follow what you trying to say - have you found any significant flaw in this scheme compared to to others or you suggesting an better solution? We can't skip hashes from the batch, because they are taken sequentially