Ingester: reject checkouts without commit_hash

[MarceloRobert]

Even though all checkouts should have a commit hash by default, it is still possible for maintainers to submit such results, and the database allows this case. The commit_hash is a very important field since it is used for accessing the treeDetails page and also navigating between commits, it should be enforced to be not null and wrong data should be rejected.

Comes from a comment on another issue, but was born from wanting group builds and tests with some key fields (#1454)

[tales-aparecida]

This needs to live in an Epic (do we use those here?) about "KCIDB cleanup"

[MarceloRobert]

I can create another issue and add this as a sub-issue, will do

[tales-aparecida]

I think commit_hash can become a required field in the schema. This would be a "breaking change", so we might want to batch "all fields that should be required" in a single KCIDB major bump

[tales-aparecida]

I think commit_hash can become a required field in the schema. This would be a "breaking change", so we might want to batch "all fields that should be required" in a single KCIDB major bump

To expand on that, I think this ticket could be moved to https://github.com/kernelci/kcidb-io/