feat(ansible): add the CARLA 0.10 Town10HD_Opt map to demo_artifacts

[hakuturu583]

Description

The Autoware maps published alongside the CARLA towns were recorded on CARLA 0.9. CARLA 0.10 re-authored those towns in Unreal Engine 5, so the published point clouds no longer describe the world a vehicle drives in, and NDT scores a confident match on geometry that has moved. AutowareFoundation/carla-ue5-maps carries point clouds recorded from 0.10 itself, one directory per world, starting with Town10HD_Opt.

This adds that download to demo_artifacts, joining the hf loop the Kashiwanoha map already runs in — as #7301 anticipated ("the next map dataset is one more loop entry"). Two changes make room for it:

• Filters are per entry. The --exclude "*.mp4" that dropped the Kashiwanoha demo video was hardcoded; entries now carry include/exclude pattern lists. This dataset drops the preview renders and every world but Town10HD_Opt.
• dest is relative to autoware_data, not maps. The dataset already namespaces its worlds under autoware_maps/, and hf download preserves the repo-relative path, so this entry points dest at maps itself and unpacks into maps/autoware_maps/Town10HD_Opt/ rather than claiming a directory of its own. That is the layout the dataset card documents, and the directory basename is the world name CARLA loads, matching the map_path convention the CARLA demo already uses.

The Kashiwanoha download is unchanged — same argv before and after.

The dataset carries no tag yet, so the revision is the commit that published Town10HD_Opt (682ddd4).

Resulting layout:

~/autoware_data/maps/autoware_maps/Town10HD_Opt
├── lanelet2_map.osm          # read by Autoware
├── map_projector_info.yaml   # read by Autoware
└── pointcloud_map.pcd        # read by Autoware, for localization

Adding a world later is one more include pattern.

How was this PR tested?

• Rendered the templated hf command through Ansible's own Jinja (ansible-playbook over the task's loop and loop_control) and compared the argv after shlex splitting. The map-carla-kashiwanoha invocation is byte-identical to main; the new one is:

  hf download AutowareFoundation/carla-ue5-maps --repo-type dataset \
    --revision 682ddd48b4cf305507646cb2f2e563bbedac4c2e \
    --include "autoware_maps/Town10HD_Opt/*" \
    --local-dir "$HOME/autoware_data/maps"

• Ran that command for real against a scratch --local-dir. It fetches exactly the three map files at the pinned revision, no preview renders:

  Fetching 3 files: 100%|██████████| 3/3 [00:04<00:00,  1.47s/it]
  
        22  maps/autoware_maps/Town10HD_Opt/map_projector_info.yaml
   2248269  maps/autoware_maps/Town10HD_Opt/lanelet2_map.osm
  22220444  maps/autoware_maps/Town10HD_Opt/pointcloud_map.pcd

  map_projector_info.yaml is projector_type: Local, and the point cloud header reports the 1 851 689 points the dataset card states.

• pre-commit run --files ansible/roles/demo_artifacts/{tasks/main.yaml,README.md} passes (yamllint, prettier, markdownlint, check-yaml).

Not exercised: a full ansible-playbook autoware.dev_env.install_dev_env --tags demo_artifacts run — demo_artifacts is never-tagged and not covered by health-check-ansible, so the role was verified at the task level as above rather than end to end.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Lb99yovCQZVXT47PYzAjXJ

[github-actions]

Thank you for contributing to the Autoware project!

🚧 If your pull request is in progress, switch it to draft mode.

Please ensure:

• You've checked our contribution guidelines.
• Your PR follows our pull request guidelines.
• All required CI checks pass before marking the PR ready for review.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c078bb7625

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[xmfcx]

I thought you were going to include kashiwanoha map in this pack. And remove the kashiwanoha entry. Did you change your mind?

[hakuturu583]

I thought you were going to include kashiwanoha map in this pack. And remove the kashiwanoha entry. Did you change your mind?

No change of mind — the plan is still to fold Kashiwanoha into carla-ue5-maps and drop the map-carla-kashiwanoha entry. It just isn’t ready yet.
What exists for the UE5 Kashiwanoha today is only the .xodr. The lanelet2 map and the point cloud recorded against the UE5 world haven’t been built yet, so there is nothing to put in carla-ue5-maps for it — a Kashiwanoha directory there would be missing exactly the two files Autoware reads for localization.
Once those are ready I’ll publish the whole set — .xodr, .fbx, lanelet2 map and point cloud — together as one Kashiwanoha entry in AutowareFoundation/carla-ue5-maps, and then remove the map-carla-kashiwanoha entry here so there is a single dataset again. Until that lands the existing entry has to stay, because the current CARLA demo (docker/examples/demos/carla/, pinned to 0.9.16) is using it.
So this PR is just the first world of the pack; the Kashiwanoha migration follows in its own PR.

[xmfcx]

I realized that this is wrong after merging it, @hakuturu583 -san could you fix it?

Suggested change

	│ ├── autoware_maps/
	│ │ └── Town10HD_Opt/
	│ ├── Town10HD_Opt