Repaired a bug causing wrong agent ids in abs_self_obs #303

	name: Continuous Integration

	on:
	push:
	branches: [main]
	pull_request:

	jobs:
	build-and-smoketest:
	runs-on: ubuntu-latest

	steps:
	- name: Checkout repository
	uses: actions/checkout@v4

	- name: Set up Docker Buildx
	uses: docker/setup-buildx-action@v2

	- name: Build Docker image
	run: docker build --tag gpudrive:latest .

	- name: Run smoke test inside Docker container (using a here document)
	run: \|
	docker run --rm gpudrive:latest /bin/bash -c "$(cat << 'EOF'
	#!/bin/bash
	set -e
	set -x
	export DEBIAN_FRONTEND=noninteractive
	echo 'Dummy cuda'
	ln -s /usr/local/cuda/lib64/stubs/libcuda.so /usr/local/cuda/lib64/stubs/libcuda.so.1
	export LD_LIBRARY_PATH=/usr/local/cuda/lib64/stubs/:$LD_LIBRARY_PATH
	echo 'Modifications to run without extra data'
	sed -i 's\|train_path: \"data/processed/training\"\|train_path: \"data/processed/examples\"\|g' examples/experimental/config/visualization_config.yaml
	sed -i '/# Load policy/{N;N;N;N;N;N;s\|# Load policy\n policy = load_policy(\n path_to_cpt=config.cpt_path,\n model_name=config.cpt_name,\n device=config.device,\n env=env,\n )\|from gpudrive.networks.late_fusion import NeuralNet\n policy = NeuralNet.from_pretrained(\"daphne-cornelisse/policy_S10_000_02_27\")\|}' examples/experimental/viz_rollouts.py
	echo 'Modifications to run without GPU'
	sed -i 's/device=\"cuda\"/device=\"cpu\"/g' gpudrive/datatypes/observation.py
	export PYTHONFAULTHANDLER=1
	ulimit -c unlimited
	echo 'Run smoketest'
	/gpudrive/.venv/bin/python examples/experimental/viz_rollouts.py
	EOF
	)"

Provide feedback