(De)serialization of Merkle Nodes

[emlautarom1]

Add a new namespace Merkle which describes Merkle constructs. For now, we describe Merkle Nodes and their corresponding (de)serialization methods. The current nodes are:

• Branch: 1 byte for type + dirty, 2 bytes for nibbles, optional 32 bytes for keccak
• Extension: 1 byte for type + dirty, var-length for path (usually it will be a nibble or a few)
• Leaf: 1 byte for type + dirty, var-length for path, 32 bytes for keccak

Test suite is included (98% coverage).

[emlautarom1]

WriteTo should have the semantics of returning the actual payload that was written (sliced if needed to make it shorter than max). WriteToWithLeftover is the one that returns what is left

All nodes now have both a WriteTo (returns the actual used bytes) and a WriteToWithLeftover (return a the leftover to be used by another writer)

[emlautarom1]

Serialization tests should assert var output = WriteTo(stackalloc ...) with X.ReadFrom(output)

All serialization tests now work like this, and they also validate that output is exactly the bytes used during WriteTo.

[Scooletz]

One comment about the leftover. Let me know what you think

[Scooletz]

If they write with Leftover it's up to the caller to get the actual payload? I'd provide just WriteTo (the usual case will be the caller serializing and playing with the payload) or the two. If needed, later we can refactor towards a struct that handles both cases;

[emlautarom1]

Included support for optional Keccaks on Branch. If no Keccak is provided, then the default (Keccak.Zero) is used. (De)serialization takes this into consideration, so the size of a Branch with no Keccak is smaller that the one with it.

[github-actions]

Package	Line Rate	Branch Rate	Health
Paprika	89%	79%	✔
Summary	89% (1754 / 1974)	79% (431 / 548)	✔

Minimum allowed line rate is 75%

[Scooletz]

LET'S GO!