feat: add content moderation module

[Deywumi-debug]

Summary

Adds a ModerationModule that scans submitted content for inappropriate material before it reaches the platform. All checks run on every POST /moderation/check request and any violation auto-rejects the content.

Changes

File	Change
src/modules/moderation/moderation.service.ts	Core moderation logic
src/modules/moderation/moderation.controller.ts	POST /moderation/check endpoint
src/modules/moderation/moderation.module.ts	Module wiring with HttpModule
src/modules/moderation/dto/moderate-content.dto.ts	Input DTO
src/modules/moderation/dto/moderation-result.dto.ts	Output DTO
src/modules/moderation/moderation.service.spec.ts	14 unit tests
src/app.module.ts	Register module behind ENABLE_MODERATION flag

Acceptance Criteria

• OpenAI moderation API — calls POST https://api.openai.com/v1/moderations via HttpService; requires OPENAI_API_KEY env var; skipped gracefully if absent or on network failure
• Profanity filtering — local regex patterns matching word stems, case-insensitive, no external dependency
• Spam detection — four signals: 10+ repeated characters, 3+ URLs, known spam phrases, >70% uppercase
• Auto-reject obvious violations — any triggered flag sets allowed: false, autoRejected: true, and a human-readable reason

API

POST /moderation/check
Content-Type: application/json

{ "content": "string (max 10,000 chars)" }

Response:

{
  "allowed": false,
  "autoRejected": true,
  "flags": ["spam", "profanity"],
  "reason": "detected as spam; contains profanity"
}

Configuration

OPENAI_API_KEY=sk-...     # optional — local checks still run without it
ENABLE_MODERATION=true    # feature flag, default: true

Testing

14 unit tests covering: clean content pass-through, profanity (case-insensitive), spam signals (repeated chars, URL flooding, spam phrases, excessive caps), OpenAI flagged/not-flagged/error/no-key, auto-reject with reason, and multiple flag accumulation.

npm test -- src/modules/moderation/moderation.service.spec.ts --no-coverage

Notes

• ModerationService is exported from the module so other modules (courses, posts, etc.) can inject it to gate content submission at the service layer.
• OpenAI failures degrade gracefully — a network error does not cause a false rejection.
• The module is feature-flagged (ENABLE_MODERATION) and can be disabled without touching application code.
  closes Add automated content safety scanning #560

[RUKAYAT-CODER]

Kindly resolve conflict and fix workflow.