test(subscription): verify cancel then re-subscribe replaces stale st…

[FaveTeamz]

…orage (#189)

• Add test_cancel_and_resubscribe_replaces_record: subscribes with amount=50/interval=600, cancels, resubscribes with amount=99/ interval=1200 and asserts new record has updated values with no stale state from the first subscription
• Assert get_subscription returns None immediately after cancel
• Add test_cancel_nonexistent_subscription_returns_error (negative case)
• Add test_double_subscribe_returns_error (guard test)
• Register contracts/subscription in workspace Cargo.toml

closes #189

[nanaf6203-bit]

Review: test(subscription): verify cancel then re-subscribe replaces stale subscription

Nice regression test - cancel-then-resubscribe is exactly the kind of state-machine corner case that gets reintroduced.

Hardening suggestions:

1. Boundary coverage. Tests rely on the now < sub.next_payment comparison inside execute_payment. There is no test for the exact boundary (now == sub.next_payment). Please add one - I want the upper-bound semantics pinned down by a test so a future refactor does not flip them silently.
2. Hardcoded ledger timestamps. Fixed sequences (100, 101, ...) are fine and deterministic, but please add a top-of-module comment mapping the sequences to intent (ledger 100 = baseline subscribe, ledger 101 = cancel, etc.) so the next contributor can decode the intent without re-tracing the diff.
3. CI status heads-up. I see 12 of 14 checks failing on this branch (build WASM, clippy, rustfmt, security audit, several cargo test runs - contracts/fractional, contracts/insurance, contracts/lending, contracts/oracle). Note that the new test is in contracts/subscription which builds green per the workspace test, so the cross-contract failures appear to be baseline issues on the branch rather than caused by this change - flagging in case you want to confirm before rebase.