Update `zktrie` Specification

[frisitano]

Describe the feature

In #36 we introduced zktrie and included the specification https://github.com/scroll-tech/reth/blob/feat/scroll-bmpt/crates/scroll/trie/assets/zktrie.md. The specification is outdated and does not include mention of domain separation for node hashing. We should update it to include this.

The constants for domain seperation in node hashing can be found here:

reth/crates/scroll/trie/src/lib.rs

Lines 16 to 29 in c960cb7

	/// The hashing domain for leaf nodes.
	pub const LEAF_NODE_DOMAIN: Fr = Fr::from_raw([4, 0, 0, 0]);
	/// The hashing domain for a branch node with two terminal children.
	pub const BRANCH_NODE_LTRT_DOMAIN: Fr = Fr::from_raw([6, 0, 0, 0]);
	/// The hashing domain for a branch node with a left terminal child and a right branch child.
	pub const BRANCH_NODE_LTRB_DOMAIN: Fr = Fr::from_raw([7, 0, 0, 0]);
	/// The hashing domain for a branch node with a left branch child and a right terminal child.
	pub const BRANCH_NODE_LBRT_DOMAIN: Fr = Fr::from_raw([8, 0, 0, 0]);
	/// The hashing domain for a branch node with two branch children.
	pub const BRANCH_NODE_LBRB_DOMAIN: Fr = Fr::from_raw([9, 0, 0, 0]);

We should also update the part which suggests that the keys are truncated to 248 bits. This is not the case, all 254 bits (bn254 field size) are used in the key space.

Additional context

No response

[MukulKolpe]

Hey @frisitano, can I work on this one?

[frisitano]

Hi @MukulKolpe, thanks for reaching out and showing interest! We are making some modifications to our implementation of the BMPT, and as such this is a moving target. I would appreciate your support on this after implementing those changes so that the spec accurately reflects what we have implemented.

[MukulKolpe]

@frisitano, Thank you for the update! Happy to assist once the modifications are in place. Let me know when we're ready to go.