Add endpoint to expose lambda init binary, increase fidelity of the kubernetes dev script

[dfangl]

Motivation

We are working on several features which need to download the init binary from LocalStack on demand.
This necessitated the inclusion of an endpoint to expose the init binary.

Also, the kubernetes dev script had some shortcomings, like the pro image not automatically selected, unable to automatically add environment variables, and not exposing a port by default.
For increased usability, I added those features.

Lastly, we now have a check if docker is available before trying to pull / inspect, to avoid needlessly verbose error messages.

Changes

• LocalStack now exposes its init binary over the <localstack_host>/_aws/lambda/init endpoint (GET)
• python -m localstack.dev.kubernetes now has new parameters and automatically determines the correct image (although the community image might still be broken and needs more fixing)
• Deploying an image lambda without the docker socket mount now only throws a single warning instead of multiple quite verbose errors.

[github-actions]

LocalStack Community integration with Pro

    2 files      2 suites   1h 41m 26s ⏱️
2 981 tests 2 675 ✅ 306 💤 0 ❌
2 983 runs  2 675 ✅ 308 💤 0 ❌

Results for commit 13c80a0.

♻️ This comment has been updated with latest results.

[simonrw]

I won't speak of the Lambda changes, but these seem reasonable, thanks!

[simonrw]

Thanks for making these improvements!

[joe4dev]

The changes look good to me, but I have not tested them. For such PRs, it would be helpful to have a "Testing" section with suggested steps to test.

The K8 dev script is great. Unfortunately, it did not work yet as expected:

• It prints "Generated file: configuration.yml" but does not actually generate a file (I had to manually copy/paste the output; maybe that's intended and the output is misleading)
• k3d cluster create ... prints INFO[0003] Prep: Network and then nothing happens 🤔 A subsequent helm upgrade ... fails with cluster unreachable.

I didn't investigate further ...

[dfangl]

1. I agree the terminology is misleading, but there is a --write flag you need to provide to write the files. I did not touch that part of the code, but good input for a follow up.
2. Not sure about that, it works fine on my linux machine. I have not tested the kubernetes dev script on macos either, however I am not sure it is related to my changes.