e2e: use stable-dind image for testing

[thaJeztah]

The edge channel is deprecated and no longer updated

[codecov-io]

Codecov Report

Merging #1992 into master will decrease coverage by <.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #1992      +/-   ##
==========================================
- Coverage   56.74%   56.74%   -0.01%     
==========================================
  Files         310      310              
  Lines       21802    21801       -1     
==========================================
- Hits        12371    12370       -1     
  Misses       8517     8517              
  Partials      914      914

[thaJeztah]

@silvin-lubecki @vdemeester PTAL; wasn't sure what we preferred here;

• use test-dind (latest "RC"); gives the advantage of testing against latest/greatest, but this image may change at points so possibly cause inconsistent results
• use stable-dind same "problem" as above
• pin to a specific version (this PR); downside is that we don't catch regressions, but positive side is that we know what we're testing against

[vdemeester]

@silvin-lubecki @vdemeester PTAL; wasn't sure what we preferred here;

* use `test-dind` (latest "RC"); gives the advantage of testing against latest/greatest, but this image may change at points so possibly cause inconsistent results

* use `stable-dind` same "problem" as above

* pin to a specific version (this PR); downside is that we don't catch regressions, but positive side is that we know what we're testing against

I think safest bet is stable-dind. We could have another job test-dind that is not a blocker (aka we would see it needs fixing but it wouldn't be required to be green to be able to merge). @thaJeztah wdyt ?

[thaJeztah]

I'm good with stable-dind (less maintenance) let me update

[thaJeztah]

updated 👍

[vdemeester]

LGTM 🐯