Interactive eval TUI should list available models by provider

[christso]

Objective

Add a provider-aware model selection step to AgentV's current interactive eval flow so users can choose a valid model for the selected target without already knowing provider-specific model strings.

Current Ground Truth

AgentV's current interactive flow is an Inquirer-based wizard that lets users:

• discover/select eval files
• choose a target
• set advanced options like workers, dry-run, and cache
• review and confirm execution

It does not currently prompt for model override or provider-specific model selection.

Problem

When a selected target/provider supports multiple models, users still need to know the correct model string ahead of time and pass it outside the interactive flow.

This causes a few issues:

• trial and error with invalid model names
• provider-specific confusion about whether a string is a model name or a reasoning preset
• a gap between target selection and model configuration in the interactive flow

Proposed behavior

When the interactive flow reaches target configuration, AgentV should:

• detect the selected provider/target type
• query or enumerate the available models for that provider when possible
• show a selectable list in the interactive flow
• still allow manual entry as a fallback
• preserve current behavior for providers where listing is unavailable

Nice-to-have

• indicate deprecated models
• indicate recommended/default models
• show reasoning-related settings separately from model selection when the provider supports them
• cache provider model lists for a short period to keep the flow responsive

Why this matters

This would make the interactive flow much more discoverable and reduce provider-specific guesswork, especially for Codex, Copilot, and similar agent providers where supported models evolve over time.