[RyanPCo]

.github/workflows/ci.yml, three jobs:

• lint: ruff check egomimic tests and ruff format --check on the whole
  repo (excluding external/), both with the pre-commit pinned ruff (0.8.6).
• unit-tests: uv sync --locked from uv.lock (fails if the lock is stale),
  then pytest tests/unit, which includes the Hydra compose+resolve test over
  every shipped config and a real hydra -m launch.
• wheel: uv build --wheel, install into a fresh venv with --no-deps, import
  from outside the checkout (contents are asserted by test_packaging.py).
• pyproject pins ruff==0.8.6 so the venv, pre-commit and CI run the same ruff
  (the venv had 0.15, which flags things 0.8.6 does not); uv.lock updated.

Baseline lint cleanup so the gate is green from day one (41 errors, ruff
--fix for 32: unsorted imports, unused imports; by hand: delete two unused
locals, split one semicolon statement, # noqa: E402 on the imports kept
mid-file with the block moved from egomimicUtils.py). No import removed had
side effects. pre-commit's ruff-format then reformatted the touched files, so
five of them (dataset_view.py, stage_processed_folder.py,
aria_to_mano_convert.py, human.py, collect_demo.py) carry whitespace-only
formatting changes beyond the lint fixes.

Whole-repo format: ruff format over every file outside external/ (32 files,
11 of them notebooks, mostly JSON re-indentation), so the format check covers
the whole tree rather than only changed files.

PR auto-review is disabled: its ANTHROPIC_API_KEY account is out of credits and
every run failed. The pull_request trigger is commented out (workflow_dispatch
keeps the file valid); uncomment it once credits are restored. The wiki tracker
is untouched.

Co-Authored-By: Claude Fable 5.1 noreply@anthropic.com
Claude-Session: https://claude.ai/code/session_01UHDiN8yfERcsiFdz4mXuZi

🤖 Generated with Claude Code

https://claude.ai/code/session_01UHDiN8yfERcsiFdz4mXuZi

[RyanPCo]

• import cloud packages lazily to support local training #631 : 2 dependent PRs (#632 , #633 )
• cache dataset resolver for faster training initialization #630
• validate episode hashes in data config before training #628
• modify pi camera map to match data #629
• add tests to ensure training works between every model and data vendor #626
• run CI tests on every PR #625 👈 (View in Graphite)
• add env variables for user-specific paths #624
• build: set pyproject and uv as the dependency managers #623
• fix(configs): fix paths for data hydra yaml configs #622
• refactor(algo): delete ACT #627 : 1 other dependent PR (#621 )
• tests: create dedicated test folder with unit and integration tests #620
• add pyright type checker #619
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.