`evalview validate` — lint test files without running them

[hidai25]

What

Add evalview validate [path] that checks test YAML/TOML files for schema errors, missing required fields, and invalid values — without calling any agent.

Why

Catching bad test files before running saves time and API costs. Useful in CI as a fast pre-check.

Scope

• New command: evalview validate [path]
• Validates each file against the Pydantic TestCase model
• Reports all errors with file name + line hint, not just the first one
• Exit code 0 = all valid, 1 = errors found

Getting Started

• TestCase model with validators: evalview/core/types.py lines 159-319
• Field validators already exist: name validation (line 263), adapter whitelist (line 284), tag normalization (line 298)
• Reference for validation command: evalview/commands/skill_cmd.py lines 42-96 (skill validate does something similar for skills)
• Loader: evalview/core/loader.py

Acceptance Criteria

• evalview validate tests/ reports all errors across files
• Shows filename and specific field errors
• evalview validate tests/ --json outputs machine-readable results
• Exit code 1 on any validation error (CI-friendly)
• Works without an API key or running agent