feat: Datadog log adapter for ingestion

[leo-aa88]

Summary

Add a Datadog ingestion path so operators can pull logs from Datadog into the same pipeline as file-based raglogs ingest (normalize → fingerprint → Postgres/pgvector).

This is the suggested next implementation focus (chosen over Loki/K8s first).

Motivation

• README Roadmap lists a Datadog adapter.
• Many teams do not have log files on disk; Datadog is a common source of truth.

Scope (proposal)

• New adapter under src/.../adapters/ (or documented subpackage) that yields ParsedLogLine like the file adapter.
• CLI entry point, e.g. raglogs ingest-datadog or raglogs ingest --source datadog with required flags/env (API key, site, query/window, service tags).
• Configuration via .env / env vars (e.g. API key, app key if needed, site, default query window).
• Document limits: rate limits, max rows per run, pagination.

Acceptance criteria

• Can ingest a bounded time window from Datadog into the existing DB schema without manual export files.
• Unit tests for adapter parsing/mapping; integration test optional (mock HTTP or recorded fixtures).

References

• Existing pattern: Adding a log source adapter in README.
• Datadog Logs API: https://docs.datadoghq.com/api/latest/logs/

Notes

• Align field mapping with README JSON aliases (timestamp, message, level, service, env, trace/request IDs, host) where Datadog attributes provide them.