feat(ai): add pyoaev support for AI adversarial exposure validation

Summary

Adds the pyoaev SDK support required by the OpenAEV AI adversarial exposure
validation domain (running adversarial attacks against LLM models / AI agents and
validating AI defenses such as LLM firewalls and guardrails).

Changes

signatures: new SignatureTypes values ai_request_marker and ai_target_endpoint.
signatures: shared deterministic per-inject canary marker helper
(pyoaev/signatures/ai_marker.py), so the AI red-team injector and the AI defense
collectors compute the same marker independently (no platform round-trip).
inject_expectation: new ai_expectations_for_source(source_id) to poll agentless
DETECTION / PREVENTION expectations used by AI defense collectors.
apis: new AiTargetManager (CRUD for AI Target assets) wired on the client.

Context

This is the dependency root of the AI adversarial exposure validation feature.
It should be merged and released before the injector (ai-redteam) and collector
(ai-guardrail) work that consumes these APIs, and it pairs with the openaev
backend endpoints (/api/injects/expectations/ai/{sourceId}, /api/ai_targets).

Test plan

python -m compileall pyoaev
Lint / type-check per repo standards (ruff, mypy)
Smoke test OpenAEV(...).ai_target and inject_expectation.ai_expectations_for_source