feat(evaluator): Set up environment for custom Gym e2e test script

[JashG]

Summary

PR 1 of 3 to add a manual end-to-end Gym Evaluation test script that enables testing and validating a built-in custom Gym environment against a Kubernetes deployment.

This PR focuses on the environment setup portion of the script. It supports an out-of-the-box custom Gym environment, and wires up support for bringing your own environment.

Changes

Built-in ASCII Tree environment

• Supports a built-in custom environment whose dependencies are packaged as a wheel. Includes all the relevant plumbing for generating and caching the required components of the custom environment.
• This PR also ships its own custom dependency, rather than pulling an external dependency. This gets packaged in the wheel and imported by the resource server.

Bring-your-own custom environment

• Supports validating and staging your own custom environment via CLI args. You are responsible for packaging external dependencies as wheels and supplying those as a wheels-v1 environment.
• Automatically extracts relevant metadata from the environment used in later PRs to validate the job outcome.

Type of Change

• Code change (feature, bug fix, or refactor)
• Code change with documentation updates
• Documentation only
• Contributor tooling or automation
• CI, build, or test infrastructure

Quality Gates

• Tests added or updated for changed behavior
• Existing tests cover changed behavior — justification:
• Tests not applicable — justification: This adds a manually invoked validation workflow; targeted linting and type checks cover the preparation code.
• Documentation updated for user-visible behavior
• Documentation not applicable — justification: This adds a manually invoked validation workflow; not published anywhere.

Verification

• Pull request title follows the repository's Conventional Commit format
• Every commit includes an appropriate Signed-off-by: trailer
• uv run pre-commit run -a passes, or any blocked checks are identified below
• Targeted tests pass, or tests are marked not applicable above
• No secrets, API keys, or credentials are included

Targeted validation:

• Targeted ruff check and ruff format --check passed for the changed Python files.
• Commit hooks passed Ruff, formatting, ty, DCO, and repository checks.
• Full pre-commit validation was blocked only by the unavailable Studio lint-staged command in the isolated worktree.

[github-actions]

Suite	Lines Covered	Line Rate	Branch Rate
Unit Tests	44046/56011	78.6%	62.2%
Integration Tests	27458/53280	51.5%	22.7%