Add split-invariance tests for unlock_assets partial withdrawals

[Smoothjane]

Summary

Every existing partial-unlock test exercises exactly one fixed split of a withdrawal. checkpoint_position runs on every unlock_assets call and folds already-accrued credits into total_credits — a genuinely different code path per partial call than a single full unlock — so whether the final outcome (token balance, credits, cleared position) is actually invariant to how a withdrawal is partitioned was asserted nowhere, only assumed from checkpoint_position's formula being linear in amount.

• test_unlock_assets_final_outcome_is_invariant_to_how_the_withdrawal_is_split: locks 1,000, advances 10 ledgers, then withdraws the full amount via 4 different partitions ([1000], [500, 500], [300, 200, 500], [1, 1, 998]) — varying both partition count and sizing per the acceptance criteria — with zero further ledger advance between calls within a partition (isolating this from farming-pool: no property-based test enforces that total accrued credits are independent of checkpoint frequency #75's already-covered time-invariance). Asserts identical final token balance, calculate_credits(), cleared position, and — since every call after the first has zero elapsed ledgers — an identical final cumulative total_credits value in the last unlock's emitted event across every partition.
• test_unlock_assets_split_across_min_lock_period_boundary_reaches_same_final_state_as_single_unlock: covers the issue's explicit edge case — one scenario unlocks everything in a single call the moment the position matures; the other attempts (and correctly gets rejected for) an early partial unlock before maturity, then completes the withdrawal split across two calls once matured. Both reach an identical final state.
• Both use the existing setup/setup_with_lock_period/advance_ledgers harness, no new test infrastructure.
• Verified the first test actually catches a regression: temporarily removed checkpoint_position's checkpoint_ledger update (a real bug class — forgetting to advance the checkpoint after banking credits) and confirmed the [500, 500] partition diverges to 15_000 instead of the invariant 10_000, while [1_000] alone stayed correct — exactly the kind of split-count-dependent bug this test exists to catch. Reverted before committing.

Test plan

• Both new tests pass; confirmed one of them genuinely fails under an injected regression (see above), not just tautologically green
• cargo test --workspace (mirroring CI's own steps) — 184/184 passing across factory, farming-pool, vesting-wallet
• cargo build --workspace --target wasm32v1-none --release — succeeds
• cargo fmt --all -- --check — clean
• cargo clippy --workspace --all-targets -- -D warnings — clean

Closes #123

[netlify]

✅ Deploy Preview for sdcontracts ready!

Name	Link
🔨 Latest commit	c7a39a1
🔍 Latest deploy log	https://app.netlify.com/projects/sdcontracts/deploys/6a84e070b954cd00095603fd
😎 Deploy Preview	https://deploy-preview-138--sdcontracts.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.
🤖 Make changes	Run an agent on this branch

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[prodbycorne]

Nice workdone