Package updates

[tjkirch]

Description of changes:

2034210b Update cni to 0.8.1
563af8e5 Update e2fsprogs to 1.46.2
af0f8eed Update ecs-agent to 1.51.0
c44642de Update kernel to 5.4.105
bd552753 Update kubernetes-1.18 to 1.18.17
e86b79a5 Update kubernetes-1.19 to 1.19.9
3faae4f5 Update libattr to 2.5.1
141354a2 Update libcap to 2.49
1218d9d5 Update libexpat to 2.3.0
b78af2f0 Update libselinux to 3.2
27bb3bb9 Update libsemanage to 3.2
25295b36 Update libsepol to 3.2
266b4184 Update policycoreutils to 3.2
91f119e3 Update systemd to 247.6
dc7f1eb3 Update wicked to 0.6.65

Not updated: libxcrypt

Testing done:

• Compared console output before and after - some minor changes from packages: change systemd boot sequence #1423 but nothing scary. Saw the isolate sequence.
• No SELinux errors / AVC denials.
• sestatus still works fine. (It moved.)
• Compared kubelet 1.19 logs - output is less consistent and harder to compare, but (expected) errors are the same, and nothing new looked bad.
• Compared kubelet 1.18 logs - same.
• Ran pod on 1.19 OK.
• Ran pod on 1.18 OK.
• Compared ecs-agent log before and after - just version and IDs different, looks good.
• Ran an ECS task OK.
• aws-dev variant OK, ran a docker container.
• aarch64 build of aws-k8s-1.19 is OK too.

Terms of contribution:

By submitting this pull request, I agree that this contribution is dual-licensed under the terms of both the Apache License, version 2.0, and the MIT license.

[tjkirch]

^ Removed the bash update from the PR. There's a build failure that's only occurring in some builds that I haven't figured out how to reproduce. It could be a timing error related to when host helper programs are built..? (builtins/builtext.h: No such file or directory)

There weren't any important changes in bash, and it's only used in the -dev variants for console login, so very low priority.

[bcressey]

I'd like to punt on the systemd update for now, but the others look good to me.

[tjkirch]

^ Removes the systemd update. Retested to confirm systemctl status was good, pod ran OK.

The hesitation here is that we had a somewhat painful issue with systemd 247.0 that led us to move to the -stable tree, i.e. 247.4, so there would be a bit more bake time before we incorporate new major versions.

However, I think I should actually update to 247.6, not stay on 247.4. Unless anyone objects, I'll do that soon.

[tjkirch]

^ Update more conservatively to systemd 247.6. Repeated basic testing OK.