feat(config): lower scenarios into tests

[christso]

Summary

AgentV can now load Promptfoo-style scenarios as a pre-expansion matrix helper: each scenario config row is lowered with each scenario test row into ordinary AgentV tests before the normal prompts x tests x targets expansion.

This keeps scenarios as authoring sugar rather than a new run artifact or grouping model. providerPromptMap and provider_prompt_map remain rejected because prompt selection should stay explicit through AgentV prompts, targets, files, tags, and CLI selection rather than Promptfoo provider identity baggage.

Validation

After PR #1691 merged to main, this branch was rebased onto origin/main and validated with:

• bun test packages/core/test/evaluation/validation/eval-file-schema.test.ts packages/core/test/evaluation/validation/eval-validator.test.ts packages/core/test/evaluation/loaders/jsonl-parser.test.ts packages/core/test/evaluation/validation/eval-schema-sync.test.ts
• bun --filter @agentv/core typecheck
• bun --filter @agentv/core lint
• bun --filter @agentv/core build

The worker also regenerated skills-data/agentv-eval-writer/references/eval.schema.json and verified the pinned Promptfoo clone at 6bfc5a0c7f16f9c4717ac731d276b578e63d0769.

Live provider dogfood is deferred to the final docs/examples pass (av-kfik.16) because this PR changes deterministic schema/validation/loading and does not alter provider/grader adapter behavior or run artifact contracts.

Related

Related: av-kfik.50
Supersedes closed stacked PR #1692.

---

[cloudflare-workers-and-pages]

Deploying agentv with    Cloudflare Pages

Latest commit:	c2d5e12
Status:	✅  Deploy successful!
Preview URL:	https://275b7db3.agentv.pages.dev
Branch Preview URL:	https://feat-av-kfik-50-scenarios-lo.agentv.pages.dev

View logs

[christso]

Coordinator review update: after rebasing onto merged main, I found and fixed one validation bug before merge. Scenario tests without their own prompts/provider_output now validate when every scenario config row supplies prompts/provider_output, matching runtime lowering. Added positive and mixed-config negative regression tests. Local focused tests now pass with 212 cases, plus core typecheck, lint, build, and git diff --check.