Summary

The shielded merkle tree test suite that consumes this package (MerkleTreeServiceTest, MerkleTreePruningTest, MerkleTreeEndToEndTest) was spending ~452s — almost entirely in WASM module setup, not actual tree operations. This PR makes those tests ~41x faster (452s → 10.9s) with three layers of optimization in wasm-privacy-coin.

Root cause

WasmBridge re-parsed the 466 KB privacy_coin.wasm binary and re-instantiated a fresh Chicory Instance on every ShieldedMerkleTree creation (~14s per cycle). Chicory 1.1.1 was interpreter-only, so even after caching, interpreted WASM execution dominated the remainder.

Changes

1. Module caching (WasmBridge.java) — the parsed WasmModule is cached in a static volatile field (immutable, safely shared across threads). Only the first WasmBridge() pays the parse cost.

2. Instance pooling (WasmBridge.java) — a ConcurrentLinkedDeque<Instance> reuses Chicory Instances. On close(), drop_tree drops the in-WASM tree and the Instance returns to the pool. The next WasmBridge() borrows it; from_frontier/from_state re-initializes the tree state in the existing linear memory. Failed init uses discard() to avoid polluting the pool with a dirty Instance.

3. JIT compiler (pom.xml + WasmBridge.java) — upgraded Chicory 1.1.1 → 1.7.5 and added the compiler module. MachineFactoryCompiler.compile(module) JIT-compiles all WASM functions to JVM bytecode once per JVM; the resulting Machine replaces the interpreter, giving native-speed execution that the JVM C2 JIT further optimizes.

ShieldedMerkleTree.close() now delegates to WasmBridge.close() (which drops the tree + pools the instance), and create() uses discard() on init failure.

Results (downstream shielded merkle tree test suite)

Verification

• All 41 wasm-privacy-coin tests pass.
• The downstream consumer's merkle tree tests pass.
• The Chicory version bump (1.1.1 → 1.7.5) is a breaking change for any consumer still on 1.1.1.

Refs: T1-3728