[Optimization] Conditional storage TTL bump – fixes #16

[onahiOMOTI]

Closes #16

Changes

• Added ensure_sufficient_ttl() helper
• Uses extend_ttl(1000, max_ttl) so the bump only happens when remaining lifetime < 1000 ledgers
• Automatically keeps both instance storage and contract WASM code alive
• Called on every public endpoint (create_grant, withdraw, set_pause)

This eliminates unnecessary gas on 99% of interactions while guaranteeing the contract never expires.

Follows current Stellar best practices (see: https://developers.stellar.org/docs/build/guides/archival/test-ttl-extension and extending-wasm-ttl guide).

Testing

• Deploy & call functions normally → gas should be lower on repeated calls
• Advance ledger in tests until TTL < 1000 → confirm extension happens

[onahiOMOTI]

Closes #17

Changes

• Added dispute_active: bool to Grant struct
• Stored global arbiter Address in instance storage (set via new set_arbiter fn)
• New set_dispute_state(grant_id, active): callable only by arbiter
• withdraw now blocked if dispute_active == true
• Future-proof for "Court" system (block other mutating actions later)

Acceptance Criteria

• Store arbiter address
• If dispute_active, block withdraw (recover can be added later)

Low priority / governance label — minimal change, no breaking behavior.

Test plan:

• Deploy, set arbiter
• Create grant
• Try withdraw normally → ok
• Arbiter calls set_dispute_state(true) → withdraw panics
• set_dispute_state(false) → withdraw works again

Ready for review!