Out-of-repo snapshot storage

[t0yv0]

Hello!

• Vote on this issue by adding a 👍 reaction
• If you want to implement this feature, comment to let us know (we'll work with you on design, scheduling, etc.)

Issue details

Currently capturing snapshots in the form of Pulumi state files or GRPC logs creates data that informs the tests, and it's expected to be checked into the repository. This may be an unfortunate choice since it takes up space and creates spurious diffs, where ideally it is managed more automatically as opaque data.

We might want to consider out-of-repo storage options:

• some place on GitHub Actions platform like artifacts
• sidecar S3 buckets
• https://www.pachyderm.com there is tooling out there in the space of checking in data descriptors but managing immutable data out-of-repo - I just name the one I remember

Whatever we design should be ideally compatible with:

• external contributors having read access to the data if it informs tests they want to run
• fully automated refresh jobs that keep the data fresh for Pulumi maintainers internally
• fully automated baseline version upgrades
• perhaps the above done in a way that's easy to replicate for external provider maintainers

For something lighter weight, we can just configure GitHub to ignore it when showing diffs.

Affected area/feature