Expose the internal state of Merkle Tree to allow more efficient sub root digest calculation

[masih]

As part of chain exchange, once we receive a chain the key for every subchain of it is calculated to speed up discovery, implemented here. The current implementation uses the naive/simple approach by calling merkle.Tree for every sub chain. This effectively re-hashes the same information over and over for smaller sub-chains.

The digest calculation can be made much cheaper if the merkle implementation exposes internal state such that the intermediate state from calculating the longest sub-chain can be used to calculate the smaller sub-chain digests.

[masih]

Discussed in standup 16-01-2025: Lower priority; stretch for M2.