On each PR to main, we tag a new auto-version and push a .deb as a github release and a python package to pypi.

If the pypi package is identical to a prior release, pypi refuses to take the new release and we end up out of sync.

For example, after merging #1500 the release to pypi fails here: https://github.com/panda-re/panda/actions/runs/9357671824/job/25758919983.

Which means our latest version is now 1.8.28 for a .deb and 1.8.27 for the pypi release.

Consumers of our package can't use logic like this when these versions are out of sync:

ARG PANDA_VERSION="1.8.28"
ARG BASE_IMAGE="ubuntu:22.04"

FROM $BASE_IMAGE
# Download and install deb
RUN wget -O /tmp/pandare.deb https://github.com/panda-re/panda/releases/download/v${PANDA_VERSION}/pandare_$(echo "$BASE_IMAGE" | awk -F':' '{print $2}').deb && apt install -f *.deb

# Download and install panda pypi
RUN pip install pandare==${PANDA_VERSION}

I think it would be fine to have these versions completely disconnected or perfectly in sync. I don't think this current mostly-in-sync approach is good.