Consensus constant for epoch height calculation does not seem correct

[Cifko]

Current epoch is currently computed as the height of the chain divided by the consensus constant

    pub fn block_height_to_epoch(&self, height: u64) -> VnEpoch {
        VnEpoch(height / self.vn_epoch_length)
    }

So if we change the consensus constant vn_epoch_length to something larger, we suddenly can get epoch number from the past
e.g.
if the current height is 1000 and the vn_epoch_length is 10, the current epoch is 100, and from 1001 new consensus constant is set with vn_epoch_length to be 20, and now the epoch is 50 (so the 2nd layer will go back in time, probably breaking itself)

Cifko PR 5749