Reduce instance-storage growth by moving per-vault records to persistent storage with TTL bumps

[1nonlypiece]

📌 Description

Every DataKey::Vault(id) record in src/lib.rs is written to
env.storage().instance(). Because instance storage is loaded and re-serialized
in full on every contract invocation, storing an unbounded, ever-growing set of
vaults there makes each call progressively more expensive and risks hitting
instance-size limits as vault_count grows.

Goal: move per-vault records to persistent storage keyed by id (keeping
only small counters like VaultCount in instance storage), and bump the
entries' TTL on access so active vaults are not archived.

🎯 Requirements and Context

• Must be tested and documented; no behavioural change to entrypoint results.
• Migrate DataKey::Vault(id) reads/writes from instance() to persistent().
• Add extend_ttl bumps on read/write so long-lived vaults survive state expiry.
• Keep VaultCount (and any future admin/pause flags) in instance storage.
• Preserve get_vault_state / vault_count semantics exactly.

🛠️ Suggested Execution

1. Fork the repo and create a branch

git checkout -b feature/persistent-vault-storage

2. Implement changes

• Contract: update all DataKey::Vault access in src/lib.rs to persistent
  storage with TTL extension; document chosen TTL constants.
• Tests: extend tests/lifecycle.rs to assert state survives ledger advancement
  and that reads/writes still return correct vaults.
• Docs: add docs/STORAGE_MODEL.md explaining instance vs persistent choices and
  the gas/size rationale; cross-link from README.md.
• Add NatSpec-style /// comments on the storage helpers.

3. Test and commit

• Run cargo test.
• Cover edge cases: many vaults created, read after large ledger jumps, TTL bump
  on each lifecycle transition, get_vault_state for missing id still None.

Example commit message

perf: move vault records to persistent storage with TTL extension

✅ Guidelines

• Minimum 95% test coverage on the migrated storage paths.
• Clear documentation of the storage model and TTL rationale.
• No regressions on existing lifecycle and read semantics.
• Timeframe: 96 hours.

🏷️ Labels

type-performance · area-contracts · type-refactor · MAYBE REWARDED ·
GRANTFOX OSS · OFFICIAL CAMPAIGN

💬 Community & Support

• Join the Disciplr contributor Discord to coordinate, ask questions, and get
  unblocked fast: https://discord.gg/xvNAvMJf
• Please introduce yourself before you start so we can avoid duplicate work,
  pair you with a reviewer, and merge your PR quickly.

[Senorespecial]

I would like to work on this issue please assign