Fix bugs in GatewayWithAttachedRoutes

[arkodg]

What type of PR is this?

What this PR does / why we need it:

• Set status to False for notAccepted Route condition

• Remove Accepted condition from status.listeners[0].conditions since it is not mandatory for the Accepted condition to be set. In this case the ResolvedRefs=False due to InvalidCertificateRef so an implementation may choose to not generate any listener config

Which issue(s) this PR fixes:

Also Fixes # #2526

Does this PR introduce a user-facing change?:

Fixed the status condition requirements in the `GatewayWithAttachedRoutes` test

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: arkodg

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• conformance/OWNERS [arkodg]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[arkodg]

cc @danehans

[youngnick]

Uh, where do we say that the Accepted condition is not mandatory? That is a mistake, it should be mandatory.

[arkodg]

@youngnick based on the definition here

gateway-api/apis/v1/gateway_types.go

Line 918 in cb82fb5

// This condition indicates that the listener is syntactically and

an implementation should set the Listener Condition to Accepted, only if it generates any valid data plane config.
For this case, since gateway-conformance-infra/does-not-exist secret does not exist, the implementation may choose to not generate any data plane config, so imo the conformance test should not enforce the Accepted=True condition

[youngnick]

I think that we may need to update that definition - as part of clarifying AttachedRoutes, we've effectively relaxed that so that Accepted means more "Accepted for processing" rather than "accepted for config".

@robscott, you need to see this one.

[youngnick]

@robscott and I talked about this, and agreed that this change is good for now, but I've also logged #2536 to come back and clear this up a bit.

[robscott]

Thanks @arkodg! Also thanks to @youngnick for all the thought put into this one!

/lgtm

[sayboras]

is it conflicting with

gateway-api/conformance/tests/httproute-invalid-backendref-nonexistent.go

Line 48 in faba258

gwAddr := kubernetes.GatewayAndHTTPRoutesMustBeAccepted(t, suite.Client, suite.TimeoutConfig, suite.ControllerName, kubernetes.NewGatewayRef(gwNN), routeNN)

?

[michaelbeaumont]

Yeah. I don't see how an implementation can pass both of these tests.
We have this resource that should be Accepted as @sayboras mentioned that's essentially identical to the resource in the test from this PR that should NOT be Accepted.

[michaelbeaumont]

Addressed in #2548