feat(task-program): configure planner from execution policy

[yuecideng]

Description

This PR adds Task Program execution-policy configuration for motion planner backends.

Execution policies can now declare a closed planner selection and typed planner options:

planner:
  type: toppra
  config: {}

The declaration is validated during configuration loading, included in the configured integration identity, and passed to the live simulation factory. Runtime-owned robot_uid, simulation instance IDs, and cuRobo collision objects are filled from the selected environment and scene. Existing deployments retain TOPPRA by default; the reference motion-generation policy now declares it explicitly.

The change keeps planner configuration in the Task Program execution policy while leaving program.yaml provider-independent and keeping physical collision objects owned by the live scene.

Validation

• black --check --diff --color ./ — passed
• python docs/scripts/check_api_docs.py — passed (2387/2387)
• python .agents/skills/project-dev-context/scripts/context.py check — passed
• python -m py_compile on changed Python files — passed
• Focused pytest collection was attempted:
  • tests/gym/envs/task_program/test_configured_integration.py
  • tests/gym/envs/task_program/test_simulation_environment.py
  • blocked during collection by the local environment (torch missing; installed SciPy/Scikit-learn binary signature mismatch), with no test assertions executed.

Dependencies: none.

[greptile-apps]

Confidence Score: 5/5

[Medium risk] Adds planner selection to execution policy configuration.

The PR appears safe to merge based on the reviewed changes and the resolved prior findings.

Summary

The PR lets execution policies select a typed motion planner, includes that selection in configured integration identity, and passes it to the live simulation factory. The latest changes make explicit null cuRobo nested options fail at the mapping check and add regression tests.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart LR
  Policy[Execution policy planner declaration] --> Decode[Validate typed options]
  Decode --> Identity[Configured integration identity]
  Decode --> Factory[Live simulation factory]
  Factory --> Runtime[Bind robot and simulation]
  Runtime --> Scene[Attach cuRobo scene objects if selected]
  Scene --> Generator[Motion generator]

Loading

_{Reviews (3) · Last reviewed commit: "fix(task-program): reject null planner s..."}