Leverage "reorg_cache" when hydrating compact blocks

[antiochp]

The reorg cache is used in the case of an actual reorg, to allow the local txpool to be put back in a consistent state. We do not currently take advantage of it in the case of a short losing fork.

Say we have [tx1] in the txpool and we accept block B (containing the kernel for tx1), this evicts tx1 from the txpool.
If we then see block B' (a fork at same height as B), also containing the kernel for tx1
we don't see tx1 in the txpool and we currently go ask a peer for the full block B'.

But - we do have tx1 in the reorg cache, so we could go look for it in there (I think this would be safe to do).

We use the reorg cache to keep the txpool consistent, we don't currently take advantage of it directly when accepting blocks.