Create devcontainer for quick dev setups

.devcontainer/Dockerfile

@@ -0,0 +1,19 @@
+FROM ubuntu
+
+ARG VENV_PATH=/opt/dev-venv
+ENV VENV_ACTIVATE=${VENV_PATH}/bin/activate
+
+RUN apt update
+RUN apt install -y python3.10-venv nvidia-cuda-toolkit gcc vim git
+
+# the mkdir command bypasses a profiler error, which allows us to run it with
+# host code only to at least check that the profiler parameters are correctly
+# provided; without this line, some tests will fail
+RUN mkdir -p /usr/lib/x86_64-linux-gnu/nsight-compute/sections
+
+# we create the virtualenv here so that the devcontainer.json setting
+# python.defaultInterpreterPath can be used to find it; if we do it in the
+# post_create.sh script, the virtualenv will not be loaded and features like
+# pylance, black, isort, etc. will not work
+RUN python3.10 -m venv ${VENV_PATH}
+RUN echo "source ${VENV_ACTIVATE}" >> ~/.bashrc