refactor: prover logic, dependency updates

[sergeytimoshin]

1. Dependency updates

• Updated light-concurrent-merkle-tree, light-batched-merkle-tree, light-merkle-tree-metadata, light-compressed-account, light-hasher, light-merkle-tree-reference) to a newer revision (a160153... from 368f9f0...).
• Removed light-sdk.

2. Public Transaction Events

• Merged PublicTransactionEventV1 and PublicTransactionEventV2 into a single PublicTransactionEvent struct.

3. Prover client

• Proof Component Types: The ProofABC struct now uses fixed-size arrays, providing stricter type definitions.
• Proof Compression and Negation:
• Replaced the local change_endianness function and refined G1 point negation in gnark.rs. The new negate_g1 function now uses convert_endianness from solana_program.
• Introduced a compress_proof function that utilizes alt_bn128_g1_compress and alt_bn128_g2_compress for standardized proof compression, replacing the previous negate_proof logic.
• Logic Adjustment: The root_seq assignment in has been adjusted to be conditional based on tree_height.

4. Error Handling

• Functions deserialize_hex_string_to_bytes, proof_from_json_struct, and get_public_input_hash now return Result types.

[ananas-block]

This looks suspicious. Are you sure this works once the tree is being forested?

[ananas-block]

My intuitions is that it should be something like:
root_seq: if TREE_HEIGHT_V1 == tree_height { (seq % 2400) } else { (seq % 20) },

[sergeytimoshin]

It's for the case when tree is empty:

photon/src/ingester/persist/persisted_indexed_merkle_tree.rs

Line 134 in 3d2dfe8

if btree.is_empty() {

For non-empty trees we use

photon/src/ingester/persist/leaf_node_proof.rs

Lines 190 to 193 in 3d2dfe8

	let root_seq = match node_to_model.get(&(leaf_node.tree.to_bytes_vec(), 1)) {
	Some(root) => root.seq,
	None => None,
	};

[ananas-block]

Then we don't need to do this:

for elem in proof.iter() {
            root = compute_parent_hash(root, elem.to_vec()).map_err(|e| {
                PhotonApiError::UnexpectedError(format!("Failed to compute hash: {}", e))
            })?;
        }

We can just do:
ZERO_BYTES[tree_height]

and there is no need to validate the Merkle proof since it's just zero values.
We can remove:
merkle_proof.validate()?;

[ananas-block]

Wdyt about extracting this branch if btree.is_empty() { into a function so that the empty case is not ambiguous?

[sergeytimoshin]

Then we don't need to do this:

for elem in proof.iter() {
            root = compute_parent_hash(root, elem.to_vec()).map_err(|e| {
                PhotonApiError::UnexpectedError(format!("Failed to compute hash: {}", e))
            })?;
        }

We can just do: ZERO_BYTES[tree_height]

We can't do that, because the zeroeth_element_hash is not ZERO_BYTES[0]. When btree.is_empty(), we are constructing a proof for the zeroeth_element.

[sergeytimoshin]

Added a unit test to confirm the distinction https://github.com/Lightprotocol/photon/blob/sergey/cleanup-deps/src/ingester/persist/persisted_indexed_merkle_tree.rs#L702-L715

[sergeytimoshin]

Wdyt about extracting this branch if btree.is_empty() { into a function so that the empty case is not ambiguous?

Sure https://github.com/Lightprotocol/photon/blob/sergey/cleanup-deps/src/ingester/persist/persisted_indexed_merkle_tree.rs#L135