Stabilize test cases which depends on execution timing

[kenhys]

Problem

sometimes GitHub Actions CI fails.

# Running:

................E.................................................

Finished in 30.258607s, 2.1812 runs/s, 4.4285 assertions/s.

  1) Error:
TestWatchNamespaces#test_0012_namespace watch retries when 410 Gone errors are encountered:
NoMethodError: undefined method `>=' for nil:NilClass
    test/plugin/test_watch_namespaces.rb:250:in `block (3 levels) in <class:TestWatchNamespaces>'
    test/plugin/test_watch_namespaces.rb:241:in `block (2 levels) in <class:TestWatchNamespaces>'
    test/plugin/test_watch_namespaces.rb:240:in `block in <class:TestWatchNamespaces>'

66 runs, 134 assertions, 0 failures, 1 errors, 0 skips
Coverage report generated for Unit Tests to /home/runner/work/fluent-plugin-kubernetes_metadata_filter/fluent-plugin-kubernetes_metadata_filter/coverage.
Line Coverage: 92.23% (962 / 1043)
rake aborted!
Command failed with status (1)
/home/runner/work/fluent-plugin-kubernetes_metadata_filter/fluent-plugin-kubernetes_metadata_filter/vendor/bundle/ruby/3.2.0/gems/rake-13.3.0/exe/rake:27:in `<top (required)>'
/opt/hostedtoolcache/Ruby/3.2.9/x64/bin/bundle:25:in `load'
/opt/hostedtoolcache/Ruby/3.2.9/x64/bin/bundle:25:in `<main>'
Tasks: TOP => test => base_test
(See full trace by running task with --trace)
Error: Process completed with exit code 1.

Expected

Always CI should pass

Additional Information

It seems that it raise Exception when timeout occurs before @stats was correctly set.
so these flaky test depends on execution timing

[Watson1978]

Currently, when 410 Gone errors occur, the system appears to retry immediately.
If this condition persists for a while, the continuous and immediate retries cause the CPU usage to spike to nearly 100%.

It might be better to introduce a wait mechanism, such as an exponential backoff, even for 410 Gone errors to prevent this high CPU load issue.