feat: migrate shield-swap-sdk to the new shield_swap.aleo stack

[iamalwaysuncomfortable]

Migrates shield-swap-sdk from the retired shield_swap_v3.aleo to the freshly deployed shield_swap.aleo stack, per the migration guide (amm-v3#62 target). Hard cutover — no state carries over.

Contract cutover

• Wire layer regenerated from deployed testnet bytecode: PROGRAM_ID = shield_swap.aleo; U256 {hi, lo} prices, MerkleProof struct, PositionNFT.withdrawal, reshaped PoolState/Slot/Tick/Position/SwapOutput, new compliance records; claim_multi_hop_output and set_token_decimals gone.
• Q128.128 math: new _MAGIC_X128 tick table + MIN/MAX_SQRT_RATIO_X128 pinned from the contract and verified against the reference implementation (f(±400000) matches exactly); u256_to_int / int_to_u256_plaintext helpers.
• Raw native amounts end to end: no dust rule, no 9-decimal normalization; SlotView.price re-derived; collect_all requests exactly the chain-reported owed amounts.
• Freezelist proofs: mint / claim_swap_output / collect carry [MerkleProof; 2] arrays — empty-tree defaults while the lists are empty, wrapper_proofs= override for later.
• Immutable withdrawal: mint(withdrawal=…) (defaults to recipient) is stored on the NFT; collect always pays it — the recipient kwarg is removed.
• Blinding: CLAIM_OR_SWAP_DOMAIN unchanged in the new bytecode; only DEFAULT_PROGRAM flips (v3 vectors kept as algorithm tests, new-program vectors pinned).

Automatic router dispatch

Every verb detects each token's wrapped/plain shape (from_wrapper_token_id chain probe, cached; transport errors propagate rather than misclassifying) and dispatches per the guide's §6 matrix:

• swap → shield_swap_router.swap_from_wrapped when the input is wrapped (funded with underlying records; deposit + burn happen in-transaction; amount_out_min > 0 enforced at prepare time)
• claim → 4-way dispatch on the finalized SwapOutput shapes (covers the plain-start/wrapped-claim asymmetry)
• mint/increase/collect → the 11 LP-router transitions, wrapper proofs interleaved per the deployed input orders
• decrease/burn → always direct

Record selection understands credits.aleo (microcredits) and skips recipient-bound wrapper records. Mirrored in AsyncShieldSwap.

Supporting changes

• aleo.codegen (main SDK): fixed-length Array ABI type support (fmt_array, toposort/ref-check recursion into array elements).
• API layer → staging: DEFAULT_API_URL and regen default now https://amm-api-staging.dev.provable.com (SHIELD_SWAP_API_URL override); models regenerated (wrapper_program → amm_token_program, new underlying_program/underlying_token_id drive record-program resolution); redeem no longer rotates the session credential.
• Devnode harness: deploys the new 5-program stack (core + multisig + freezelist + 2 test tokens), initializes the freezelist, registers plain tokens via the two-arg allow_token with idempotent bootstrap.

Verification

• shield-swap-sdk unit suite: 171 passed
• main SDK suite: 905 passed (includes codegen Array coverage)
• live read + drift tier (real testnet + staging API): 11 passed — pinned ABI matches deployed bytecode, X128 invariants on live slots, local pool/tick key derivation parity, registry↔chain wrapped-detection agreement
• devnode lifecycle (hermetic, full stack): 11 passed — pools → mint → increase → decrease → both-direction swaps + claims → owner≠withdrawal collect payout → burn
• live write tiers (plain + wrapped swap roundtrips) are wired and skip cleanly pending DPS e2e credentials