feat(agents): expose draft readiness evaluation

[morgmart]

Summary

• Adds one backend-owned command for evaluating either a complete unsaved new-agent configuration or a patch to one exact saved agent.
• Projects drafts through the same create/update helpers and fallible effective-harness resolver used by Save and Start, then runs the existing Rust readiness engine.
• Returns presentation-ready ready plus structured Requirement rows with the existing surface discriminator.
• Releases the managed-agent store lock before readiness probes and routes CLI login checks through the existing bounded process runner.
• Adds TypeScript API types/wrapper and reuses the shared requirement type for config nudges.

Smallest truthful backend contract

This uses one dedicated Tauri command: evaluate_agent_readiness_draft.

The request explicitly distinguishes:

• new: a complete standalone configuration using the same authoritative agentCommand field as Create;
• existing: an exact saved-agent identity plus the same patch semantics as Update.

The command does not implement a second readiness or configuration model. Create and preview share provider/model normalization, edit and preview share update helpers, and both flow through the same fallible effective-config and harness resolution used by Start before reaching agent_readiness.

Enabling contract / deferred work

This is an enabling backend contract for the new desktop client. It intentionally does not migrate or delete the existing frontend provider/form metadata table in this PR. Consumer migration remains a follow-up once the replacement client consumes this command. The legacy runtime-status local_setup boolean also remains unchanged for compatibility.

Tests

• cargo test --manifest-path desktop/src-tauri/Cargo.toml commands::agent_config --lib
• cargo test --manifest-path desktop/src-tauri/Cargo.toml managed_agents::types::tests::update_request --lib
• cargo test --manifest-path desktop/src-tauri/Cargo.toml managed_agents::readiness::cli_probe --lib
• cargo test --manifest-path desktop/src-tauri/Cargo.toml managed_agents::readiness --lib
• cargo check --manifest-path desktop/src-tauri/Cargo.toml --lib
• cargo test --manifest-path desktop/src-tauri/Cargo.toml --features mesh-llm commands::agents::tests::created_inference_config_defaults_relay_mesh_model --lib
• cd desktop && node --import ./test-loader.mjs --experimental-strip-types --test src/shared/api/tauri.test.mjs src/shared/lib/configNudge.test.mjs
• cd desktop && pnpm typecheck
• just file-size-check
• Push preflight passed: desktop-check, desktop-typecheck, desktop-test, desktop-tauri-checks, file-size-check.

[github-actions]

🔐 Codex Security Review

Status: review required for the current range.

The current range is 3c7f288c60d67df78577b237e27c3dfc8831aaa1...bb5a2c4afc139095cf9d10524077f916116d740c.
A new review must complete for this exact range. When manual authorization
is required, a Block organization member must comment exactly
@buzz-security-review bb5a2c4afc139095cf9d10524077f916116d740c to authorize a new review.
Any previous review applies only to its recorded range.