fix(workflow): validate workflow_id and confine workflow files to the workflow directory

[yonib05]

Description

The workflow tool uses workflow_id directly to build the workflow file name in WORKFLOW_DIR (WORKFLOW_DIR / f"{workflow_id}.json") across the load, store, and delete operations. The only existing check is if not workflow_id (which generates a UUID for empty create ids), so a workflow_id containing directory separators or relative segments points the read/write/delete at a file outside WORKFLOW_DIR.

This change adds two layers:

1. Allowlist validation at the tool boundary: workflow_id must match ^[A-Za-z0-9._-]{1,128}$ and may not be . or ... Invalid ids return an error result. The id is validated after the create-time UUID fallback, so auto-generated ids still pass.
2. A defense-in-depth helper, resolve_workflow_path, used by load_workflow, store_workflow, and delete_workflow. It resolves the target path and raises ValueError unless the result is inside the resolved WORKFLOW_DIR, so every filesystem operation stays confined regardless of how the id reaches it.

Related Issues

Type of Change

Bug fix

Testing

Added a TestWorkflowIdValidation suite covering the allowlist (accepted and rejected ids including ../../etc/passwd, ., .., separators, and overlong ids), the path resolver, and runtime confinement of all three sinks. Ran the workflow test module (54 passed) and ruff format --check / ruff check on the changed files.

• I ran hatch run prepare

Checklist

• I have added any necessary tests that prove my fix is effective
• My changes generate no new warnings