Do not throw error when original image was missing when committing a running container

[cardyok]

What is the problem you're trying to solve

I'm not sure whether this is appropriate or not.

When image of a running container is deleted, should this container still be available for commit?

I think in current containerd design this is doable in two ways.

1. reconstruct the content, create diff, ingest into content store and recreate layers for the newly committed image (history of original image will be lost)
2. if image is missing, ask user to provide credentials for original image repo and refetch the missing layers.

there are also similar cases when image is not deleted while contents of some layers are missing (ctr image check image is incomplete), these cases can also be covered.

Describe the solution you'd like

Some feedback about this feature.

Additional context

No response

[AkihiroSuda]

reconstruct the content, create diff, ingest into content store and recreate layers for the newly committed image (history of original image will be lost)

SGTM

[cardyok]

reconstruct the content, create diff, ingest into content store and recreate layers for the newly committed image (history of original image will be lost)

SGTM

Okk, will work on the implementation. thx for replying :)