feat: add external call return-value and returndata safety analysis

[tunmisedasa]

Summary

• Implements production-grade external call return-value and returndata safety analysis with 16 rules (CP-RTD-001 through CP-RTD-016)
• Models call, delegatecall, staticcall, send, transfer, and interface-call success/returndata flows with guard and wrapper recognition
• Merges equivalent Slither findings while preserving richer ChainProof evidence paths

Architecture

Core logic in packages/core/src/returndata/:

• Call classifier identifies low-level call kinds requiring success checks
• Guard detector recognizes SafeERC20, Address.functionCall, try/catch, assembly bounds, and documented optional calls
• Decode analyzer flags unsafe abi.decode and stale returndata reuse across sequential calls
• 16 rules cover ignored success flags, overwritten results, unchecked token returns, batch failures, assembly copies, swallowed try/catch, multicall partial failures, and proxy decode assumptions

Precision/Recall

• Distinguishes intentionally optional calls via @dev optional documentation and naming conventions
• SafeERC20/Address wrapper patterns suppress token and low-level call findings
• Vulnerable fixture triggers CP-RTD-003, CP-RTD-004, CP-RTD-005, CP-RTD-007

Test plan

• npm run build
• npm run test:ci --workspace=packages/core (337 tests pass)
• Returndata analyzer unit tests with vulnerable/secure fixtures
• CLI chainproof returndata command registration
• Manual scan of examples/contracts/returndata/ via chainproof scan

Security boundaries

• Static analysis only; no runtime returndata simulation
• Slither merge is opt-in via configuration
• Bounded parsing with recursion limits on AST traversal
  closes Implement External Call Return-Value and Returndata Safety Analysis #81

[Nanle-code]

@tunmisedasa Fix conflicts

[Manuelshub]

LGTM @tunmisedasa , nice work