[chore] cache docker images

[atoulme]

Try introducing a docker cache so we avoid hitting limits on Docker Hub.

Example:

https://github.com/open-telemetry/opentelemetry-collector-contrib/actions/runs/13730940591/job/38407799262?pr=38466

=== FAIL: . TestObserverEmitsEndpointsIntegration (0.39s)
XXX: too many requests: &{AuthConfigs:map[[https://index.docker.io/v1/:{Username:](https://index.docker.io/v1/:%7BUsername:) Password: Auth:Z2l0aHViYWN0aW9uczozZDY0NzJiOS0zZDQ5LTRkMTctOWZjOS05MGQyNDI1ODA0M2I= Email: ServerAddress: IdentityToken: RegistryToken:}] HTTPHeaders:map[] PsFormat: ImagesFormat: NetworksFormat: PluginsFormat: VolumesFormat: StatsFormat: DetachKeys: CredentialsStore: CredentialHelpers:map[] Filename: ServiceInspectFormat: ServicesFormat: TasksFormat: SecretFormat: ConfigFormat: NodesFormat: PruneFilters:[] Proxies:map[] Experimental: StackOrchestrator: Kubernetes:<nil> CurrentContext: CLIPluginsExtraDirs:[] Aliases:map[]}    integration_test.go:40: 
        	Error Trace:	/home/runner/work/opentelemetry-collector-contrib/opentelemetry-collector-contrib/extension/observer/dockerobserver/integration_test.go:40
        	Error:      	Received unexpected error:
        	            	create container: Error response from daemon: toomanyrequests: You have reached your unauthenticated pull rate limit. https://www.docker.com/increase-rate-limit
        	Test:       	TestObserverEmitsEndpointsIntegration

[atoulme]

This works:
https://github.com/open-telemetry/opentelemetry-collector-contrib/actions/runs/13731289975/job/38408718503?pr=38467

Post job cleanup.
Listing Docker images.
docker image list --format '{{ if ne .Repository "<none>" }}{{ .Repository }}{{ if ne .Tag "<none>" }}:{{ .Tag }}{{ end }}{{ else }}{{ .ID }}{{ end }}'
postgres:14
testcontainers/ryuk:0.11.0
nginx:1.17
Images present before restore step will be skipped; only new images will be saved.
docker save --output ~/.docker-images.tar postgres:14 testcontainers/ryuk:0.11.0 nginx:1.17
/usr/bin/tar --posix -cf cache.tzst --exclude cache.tzst -P -C /home/runner/work/opentelemetry-collector-contrib/opentelemetry-collector-contrib --files-from manifest.txt --use-compress-program zstdmt
Cache Size: ~175 MB (18[3](https://github.com/open-telemetry/opentelemetry-collector-contrib/actions/runs/13731289975/job/38408718503?pr=38467#step:9:3)167444 B)
Cache saved successfully

[atoulme]

On subsequent builds, I see the cache being picked up. Here is a rerun of the same build:
https://github.com/open-telemetry/opentelemetry-collector-contrib/actions/runs/13731289975/job/38409134055?pr=38467

Cache restored successfully
docker load --input ~/.docker-images.tar
Loaded image: postgres:14
Loaded image: testcontainers/ryuk:0.11.0
Loaded image: nginx:1.17