perf: pre-initialized Blake2b cloning for hash hot paths

[wbbradley]

Description

Every leaf_hash and inner_hash call creates a fresh Blake2b256 hasher through new_with_params, which runs IV setup and parameter validation. With ~3M hash calls per 32 MiB encode (1M leaf + ~2M inner), this accounts for 1.1% of self-time in profiling.

This change bypasses HashFunctionWrapper in hot paths by working with blake2::Blake2b<U32> directly. Two static LazyLock hashers are pre-initialized with the leaf/inner prefix already fed in. Each hash call clones the pre-initialized state (~200 byte memcpy) instead of running full construction.

Adds leaf_hash_blake2b256 and inner_hash_blake2b256 as specialized fast-path functions, and MerkleTree::build_from_leaf_hashes_fast for Merkle tree construction. The generic leaf_hash<T> and inner_hash<T> remain unchanged for non-hot-path usage (proofs, generic tree builds).

Test plan

• CI Pipeline
• Benchmarked with profile_encoding; results are within noise of the previous commit, consistent with the small (~1%) targeted overhead.

Release notes

• Storage node:
• Aggregator:
• Publisher:
• CLI: