listener: 2nd attempt create per network filter chain stats scope

[lambdai]

Commit Message:
This is the 2nd attempt of #17931
The goal is to fix some stats leak described in #17690

The above PR is reverted because the shared grpc logger can reference a destroyed stat scope and that behavior is amplified by the above PR.
The grpc logger issue is fixed by #18067

In case further improper behaviors are detected at a network filter or any extension,
envoy.reloadable_features.per_network_filter_chain_factory_context is added here so we don't have to revert this PR.

Signed-off-by: Yuchen Dai silentdai@gmail.com

Additional Description:
Risk Level: Mid
Testing: Tests and runtime flag behavior test.
Docs Changes:
Release Notes:
Platform Specific Features:
Runtime guard: envoy.reloadable_features.per_network_filter_chain_factory_context
[Optional Fixes #Issue]
[Optional Fixes commit #PR or SHA]
[Optional Deprecated:]
[Optional API Considerations:]

[davinci26]

/assign @envoyproxy/first-pass-reviewers

[repokitteh-read-only]

@envoyproxy/first-pass-reviewers cannot be assigned to this issue.

🐱

Caused by: a #18564 (comment) was created by @davinci26.

see: more, trace.

[mattklein123]

Awesome thanks. TBH I would skip the runtime flag here, as if people see crashes we are just going to revert this anyway and try again. I don't think it's worth adding the overhead for such a small change. Can we remove that part and just revert the revert cleanly and see if it sticks?

/wait

[lambdai]

if people see crashes we are just going to revert this anyway and try again.

Thank you for the feedback. I am not sure if I understand the piece. Isn't the goal of that runtime is to avoid revert, instead, we disable the runtime feature at a smaller penalty than revert.

I am fine and be prepared to revert though.

[mattklein123]

Thank you for the feedback. I am not sure if I understand the piece. Isn't the goal of that runtime is to avoid revert, instead, we disable the runtime feature at a smaller penalty than revert.

Yes, in general, but we don't typically do it for small changes like this. More like large changes where we might not know the implications for a long time, might be difficult to revert, etc. For this, I think we are going to know pretty quickly whether it works or not, so I would recommend just trying the fix again directly without the cleanup overhead?

[lambdai]

I see. Make sense to me. Thank you for the explanation.

I clean up the runtime related code and test. PTAL

[lambdai]

/retest

[repokitteh-read-only]

Retrying Azure Pipelines:
Check envoy-presubmit isn't fully completed, but will still attempt retrying.
Retried failed jobs in: envoy-presubmit

🐱

Caused by: a #18564 (comment) was created by @lambdai.

see: more, trace.

[lambdai]

/retest

[repokitteh-read-only]

Retrying Azure Pipelines:
Retried failed jobs in: envoy-presubmit

🐱

Caused by: a #18564 (comment) was created by @lambdai.

see: more, trace.

[jmarantz]

/wait

[jmarantz]

/assign-from @envoyproxy/senior-maintainers

[repokitteh-read-only]

@envoyproxy/senior-maintainers assignee is @mattklein123

🐱

Caused by: a #18564 (review) was submitted by @jmarantz.

see: more, trace.

[mattklein123]

Thanks!