docs/examples: add multi-model x multi-metric x variability benchmark showcase

[christso]

Canonical Plan

This issue body is the source of truth for implementation.

Objective

We create one canonical showcase that demonstrates multi-model × multi-metric × variability workflow end-to-end.

Why This Matters

• The capabilities exist, but users currently assemble the workflow from multiple separate examples.
• One integrated showcase lowers onboarding time and reduces misconfiguration risk.
• It makes AgentV’s benchmark story legible and repeatable.

Why This Location

• This is a composition and communication problem, not a runtime primitive gap.
• A docs/examples deliverable can prove workflow value without core expansion.
• Keeping this in showcase/docs preserves architecture boundaries.

Architecture Boundary

Docs/examples only.

Deliverable Location

• Showcase root: examples/showcase/multi-model-benchmark/
• Suggested structure:
  • examples/showcase/multi-model-benchmark/evals/
  • examples/showcase/multi-model-benchmark/README.md
• Docs links/mentions: apps/web/src/content/docs/

Design Latitude

We choose the final scenario/domain, provider mix, and narrative flow to maximize clarity and practical utility.

Acceptance Signals

• We add a new showcase that composes targets matrix, weighted metrics, trials, and comparison workflow.
• We include low-cost/default-safe execution guidance.
• We link it from relevant docs navigation.

Non-Goals

• No new runtime features created only for the showcase.