Add Dockerfile, use poetry for dependency management and update README

[evanjt]

This PR adds a Dockerfile that will build the dependencies and obtain the checkpoints. The execution of the docker image can then operate just as the user would interact with the CLI tool:

docker run \
    -v ./example_data/input_videos:/input \
    -v ./output:/output \
    -v ./tmp:/tmp \
    deepreefmap \
    --input_video=/input/GX_SINGLE_VIDEO.MP4 \
    --timestamp=0-100 \
    --out_dir=/output \
    --fps=10 \
    --tmp_dir=/tmp

The requirements.txt has been replaced with a poetry environment in the pyproject.toml, and the README has been updated to reflect the changes.

Also included is a github action that will build the docker image on any new version tagged on the repository, and store this in the packages for this repository.