fix: integration-test cleanup

[ericzzzzzzz]

Fixes: #8116

Description

• change cleanup(delete) call in test to be run in the same process as test itself instead of running in background process, to avoid background process being killed before skaffold delete gets called.
• refactor file sync test to use different pod for each test to avoid test contamination.

[codecov]

Codecov Report

Merging #8118 (708c969) into main (290280e) will decrease coverage by 4.19%.
The diff coverage is 53.26%.

❗ Current head 708c969 differs from pull request most recent head 23be8cf. Consider uploading reports for the commit 23be8cf to get more accurate results

@@            Coverage Diff             @@
##             main    #8118      +/-   ##
==========================================
- Coverage   70.48%   66.29%   -4.20%     
==========================================
  Files         515      599      +84     
  Lines       23150    29347    +6197     
==========================================
+ Hits        16317    19455    +3138     
- Misses       5776     8447    +2671     
- Partials     1057     1445     +388     

Impacted Files	Coverage Δ
cmd/skaffold/app/cmd/completion.go	13.04% <0.00%> (-1.25%)	⬇️
cmd/skaffold/app/cmd/credits/export.go	0.00% <0.00%> (ø)
cmd/skaffold/app/cmd/deploy.go	40.90% <0.00%> (-12.94%)	⬇️
cmd/skaffold/app/cmd/test.go	44.44% <0.00%> (ø)
cmd/skaffold/app/exitcode.go	100.00% <ø> (+6.66%)	⬆️
cmd/skaffold/skaffold.go	0.00% <0.00%> (ø)
cmd/skaffold/app/cmd/inspect_tests.go	62.50% <14.28%> (-1.14%)	⬇️
cmd/skaffold/app/cmd/render.go	35.48% <18.18%> (-5.90%)	⬇️
cmd/skaffold/app/cmd/lsp.go	28.12% <28.12%> (ø)
cmd/skaffold/app/cmd/run.go	64.28% <33.33%> (-9.63%)	⬇️
... and 394 more

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[ericzzzzzzz]

hmm... This is a little awkward, TestDevSyncDefaultNamespace whole test was passing due to file-sync pod in the first test case not being deleted. We're seeing the second test case failing, this is caused by the second test case the background dev process sees the terminating status of the pod from the first test case and before that client.WaitForPodsReady("test-file-sync") has been already executed, it saw a running pod, that's due to kubeclt delete command returns after it sends request to the cluster, it doesn't wait for the deletion. Perhaps, we shouldn't share the test resource for this whole test.

[aaron-prindle]

Nice, LGTM!