feat(core): rich scenario assertions — 8 assertion types + evaluateScenario()

[sepehr-safari]

Summary

Implements 8 declarative assertion types for scenario evaluation:

1. event_order — verify events appear in a specific order
2. event_count — verify count of events (optionally filtered by action)
3. payload_field — verify a payload field value (equals/contains)
4. timing — verify timing gaps between two actions (minGapMs/maxGapMs)
5. session_state — verify session status (active/completed/aborted)
6. failure_severity — verify a specific failure has expected severity
7. no_failures — verify no failures detected
8. failure_count — verify count of detected failures (optionally by code)

New API

export function runAssertions(ctx: AssertionContext, assertions: ScenarioAssertion[]): AssertionResult[];
export function evaluateScenario(scenario: Scenario): ScenarioEvalResult;

Backward Compatibility

• Scenario type extended with optional assertions field
• Existing scenarios with only expectedFailures work unchanged
• expectedFailures comparison still evaluated in evaluateScenario()

Files Changed

• packages/toolkit/src/core/types.ts — ScenarioAssertion, AssertionResult, ScenarioEvalResult types + optional assertions on Scenario
• packages/toolkit/src/core/assertions.ts — new module (evaluators + runAssertions + evaluateScenario)
• packages/toolkit/src/core/assertions.test.ts — 27 unit tests
• packages/toolkit/src/core/index.ts — export assertions API
• packages/toolkit/src/index.ts — re-export from root barrel
• .changeset/v03-assertions.md — minor bump changeset
• CURRENT_STATE.md — updated with v0.3.0 progress

Closes #86

Checklist

• 8 assertion types implemented
• runAssertions() evaluates each assertion against analysis results
• evaluateScenario() combines parsing + detection + assertions
• Backward compatible — existing scenarios work unchanged
• Unit tests for each assertion type (pass + fail cases)
• Unit tests for evaluateScenario() with mixed assertions
• Exported from @ocpp-debugkit/toolkit/core and root barrel
• Changeset added
• CURRENT_STATE.md updated
• Lint, typecheck, test, format all pass (349 tests)