Feature/sc 005 verification aggregation engine

[Tobi-8]

Summary

• Implemented a deterministic weighted aggregation engine (VerificationAggregator.sol) that transforms individual verifications into a single canonical AggregationResult.
• Added ClaimOutcome enum (VERIFIED_TRUE, VERIFIED_FALSE, INCONCLUSIVE) and integer-only weight/confidence math in basis points (0–10 000).
• Aggregation is order-independent and reproducible: trueWeight/falseWeight are pure accumulators over frozen effectiveStake values, so every node computes the identical outcome.
• Deterministic tie handling returns INCONCLUSIVE (no randomness), and configurable minimum participation thresholds (verification count, total weight, confidence bps) are enforced.
• Exposed aggregateClaim(), getAggregation(), calculateConfidence(), calculateWeights(), and emitted ClaimAggregated events for indexers.
• Added an IVerificationSource interface (implemented by TruthBountyWeighted) so future weighting models (reputation/Sybil multipliers, quadratic voting, committee voting) can be layered without storage changes.
• Added a compatibility surface on TruthBountyWeighted (getVerificationWeight, getVerificationSupport, getClaimVerificationWindowEnd, getClaimSubmitter) for the aggregation source interface.
• Aligned the existing test suite (fuzz, invariant, simulation, upgrade) with current contract behaviour and fixed CI config (submodules, remappings, match paths) so the full pipeline passes.
  Testing
• 27 aggregation tests: unanimous/mixed/weighted outcomes, confidence calculation, tie cases, determinism (re-aggregation + order independence), stress (50 verifications), and event emission — all pass.
• Full Foundry suite: 296 tests pass (unit, fuzz, invariant).
• Full Hardhat suite: 868 tests pass.
• Gas benchmarks for the aggregation path are documented in docs/verification-aggregator-gas.md.

closes #285