volumemgr/updatestatus: fix crash by checking status.Blobs is not empty

[rouming]

There is a crash with the following backtrace:

  github.com/lf-edge/eve/pkg/pillar/cmd/volumemgr.doUpdateContentTree(0xc000cf52c0, 0xc001a8d360, 0x42)
  pillar/cmd/volumemgr/updatestatus.go:233 +0x54b6
  github.com/lf-edge/eve/pkg/pillar/cmd/volumemgr.updateStatusByDatastore(0xc000cf52c0, 0xb42d8eee7664805, 0x23f4997dd60c22a7, 0xc001a4c190, 0xb, 0xc0018c30b0, 0x2c, 0x0, 0x0, 0x0, ...)
  /pillar/cmd/volumemgr/updatestatus.go:778 +0x3d2
  github.com/lf-edge/eve/pkg/pillar/cmd/volumemgr.handleDatastoreConfigImpl(0x23890e0, 0xc000cf52c0, 0xc0015b3ce0, 0x24, 0x26f6cc0, 0xc001a5e6e0
  /pillar/cmd/volumemgr/handledatastore.go:26 +0x198
  github.com/lf-edge/eve/pkg/pillar/cmd/volumemgr.handleDatastoreConfigCreate(0x23890e0, 0xc000cf52c0, 0xc0015b3ce0, 0x24, 0x26f6cc0, 0xc001a5e6e0)
  /pillar/cmd/volumemgr/handledatastore.go:12 +0x5d
  github.com/lf-edge/eve/pkg/pillar/pubsub.handleModify(0x26ac320, 0xc00078c100, 0xc0015b3ce0, 0x24, 0xc0012d42c0, 0x2a7, 0x2a9)
  /pillar/pubsub/subscribe.go:268 +0x7fd
  github.com/lf-edge/eve/pkg/pillar/pubsub.(*SubscriptionImpl).ProcessChange(0xc00078c100, 0x3, 0xc0015b3ce0, 0x24, 0xc0012d42c0, 0x2a7, 0x2a9)
  /pillar/pubsub/subscribe.go:132 +0x586

Patch introduces a simple check before array access and if array is empty - return an error.

!!! This patch more likely fixes the effect, not the cause. I need help from you @giggsoff @milan-zededa guys, if you know this particular code path.

UPDATE: @giggsoff pointed out that the following config (originally used by the customer) does not have a proper SHA256 for the image:

        {
            "uuid": "35ce9d94-f8c8-4a76-92ad-d43d764110a2",
            "dsId": "054866e7-eed8-420b-a722-0cd67d99f423",
            "URL": "container-disk-1G.qcow2",
            "iformat": "QCOW2",
            "sha256": "",
            "maxSizeBytes": "0",
            "siginfo": {
                "intercertsurl": "",
                "signercerturl": "",
                "signature": ""
            },
            "displayName": "container-disk-1G",
            "generationCount": "0"
        },

which causes the crash (reproduced by @giggsoff). Not clear how it is possible to create such a config without SHA256 on zedcloud, but this is another question.

For now I will keep the check introduced in this PR. There are a couple of suggestions from @giggsoff for additional checks, so will think how to add those too.

[eriknordmark]

LGTM, but it presumably also makes sense to check for a sha256 earlier (in a separate PR; such a check would need to handle that a container reference might not have a sha and volumemgr uses downloading to resolve the OCI tag (e.g., latest) to get a sha.)
But let's merge the safety check in this PR.

[rouming]

Then I will keep is as a "hot fix" as is. Other checks will follow in another PR.

[rouming]

Difference to the previous version:

• add one commit with a sha256 check

[giggsoff]

It shows me the correct error when I create volume with no sha256 defined, thank you for finding and fixing it.

[rouming]

Difference to the previous version:

• commit message tweaks