Sparse checkout for git pulls

[cpinflux]

Describe the current behavior

When cloning a git repository, the entire repository is cloned, this can take longer than the flow run itself.

Describe the proposed behavior

Additional configuration in prefect.yaml, and equivalent commands should permit a sparse checkout, maybe just of a single directory.

An additional touch would be to make the selected directory become the base of the clone, rather than retaining the full path, making it easy to just keep paths relative in code.

Example Use

in prefect.yaml

pull:
- prefect.deployments.steps.git_clone:
    repository: https://github.com/org/repo.git
    access_token: '{{ prefect.blocks.secret.github-token }}'
    directory: that/one/directory/where/my/code/is

Additional context

No response

[cicdw]

Makes sense, and seems potentially doable using something like https://git-scm.com/docs/git-sparse-checkout

[tetracionist]

I had a go at creating a branch for this feature: https://github.com/tetracionist/prefect/tree/sparse-checkout-for-git-pulls
I haven't contributed to Prefect before, so I appreciate this will need more work before I create a PR for it.

It uses sparse-checkout, but to get it to work, I needed to update the Git version on my laptop to version >=2.25.0.
On the docker image for prefect-3 latest, this should be fine as it uses debian/git 1:2.39.5-0+deb12u1

I tested this using a public repository and a process worker and it gave me a directory with the folders I specified.
I'll add some tests and happy to provide any screenshots needed.

The config works like this:

pull:
- prefect.deployments.steps.git_clone:
    repository: https://github.com/tetracionist/prefect.git
    branch: sparse-checkout-for-git-pulls
    access_token:
    directories: [src/integrations/prefect-azure, src/integrations/prefect-dask]
    cone_mode: True # set to true by default, only need to include if False

Sparsly checked out repo using my fork: