feat(cli): simplify eval output surface

[christso]

Summary

Simplifies the agentv eval output surface around one canonical run directory:

• --output <dir> / config output.dir writes the canonical run directory, always including <dir>/index.jsonl plus run/test artifacts.
• --export <file> writes optional flat copies such as JSONL, JUnit XML, YAML, or HTML without changing the canonical run directory.
• Removed --out, --artifacts, --output-format, and config output.format with migration errors instead of silent reinterpretation.
• Preserves -o as the short alias for --output <dir> and rejects file-like --output results.xml with guidance to use --export results.xml.
• Keeps feat(cli): rerun captured task bundles #1335 rerun-captured-task-bundle behavior by preserving source metadata propagation during the post-feat(cli): rerun captured task bundles #1335 rebase.

Migration Note

Breaking prerelease cleanup: callers using the old flat/result flags should migrate as follows:

• --out results.jsonl -> --output <run-dir> --export results.jsonl
• --artifacts artifacts -> --output artifacts
• -o junit.xml --artifacts artifacts -> --output artifacts --export junit.xml
• --output-format jsonl|yaml|html -> remove the flag; the canonical run directory always writes index.jsonl, and extra formats use --export <file>
• config output.format -> remove it; use config output.dir only for the run directory

Known consumer audit:

• WiseTechGlobal/sdd used --artifacts; migration is --output <dir>.
• WiseTechGlobal/WTG.AI.Prompts already uses --output .agentv/results/artifacts and remains aligned with the canonical directory contract.

Rebase Status

After #1335 merged, this branch initially appeared clean only because the local origin/main was stale at 35263cd707a9a89c95728ae86beb7271b76f2358. git fetch origin --prune updated origin/main to f1162312cb7aa645653b51756acfbbed42426929, and this PR was rebased onto that commit.

Conflict resolution:

• Resolved apps/cli/src/commands/eval/run-eval.ts by keeping feat(cli): rerun captured task bundles #1335 sourceMetadataByEvalFile / rerun task-bundle metadata behavior and feat(cli): simplify eval output surface #1336 output-surface migration validation.

Verification

Post-#1335 rebase verification on head 14352fd914f24e9c4f4450ed9681472b4bd7e9f8:

• bun run build
• bun run typecheck
• bun run lint
• bun run validate:examples
• bun test apps/cli/test/eval.integration.test.ts packages/core/test/evaluation/config.test.ts apps/cli/test/commands/eval/artifact-writer.test.ts apps/cli/test/commands/eval/task-bundle.test.ts apps/cli/test/commands/runs/rerun.test.ts
• CLI smoke:
  • --out exits 1 with migration guidance
  • --artifacts exits 1 with migration guidance
  • --output-format exits 1 with migration guidance
  • file-like --output results.xml exits 1 with --export guidance
  • --output <dir> --export export.jsonl exits 0 and writes <dir>/index.jsonl plus the flat export
• git diff --check origin/main...HEAD
• Public diff scan for private/local path strings was clean outside Beads; Beads delta is limited to this Bead and follow-up av-33j.

[cloudflare-workers-and-pages]

Deploying agentv with    Cloudflare Pages

Latest commit:	14352fd
Status:	✅  Deploy successful!
Preview URL:	https://8ea05d83.agentv.pages.dev
Branch Preview URL:	https://feat-av-eval-output-config-s.agentv.pages.dev

View logs