Extend platform acceptance for Release 2.0.0

Extends the existing r2_existing_cluster scenario with one disposable k3d runner for local use and GitLab CI.

• Container: onboarding, authorization, common request failures, zones, deployment, CPU/memory requests, observed endpoint, HTTP Pod identity and termination.
• HELM: the same lifecycle using a served chart archive. Resources remain chart-owned; CAMARA compute-to-values mapping is outside this proof.
• EURECOM NEF/CoreSim: actual simulated UE, four configured QoD profiles, refusals and diagnostics, location, and device TI toward the deployed application host IPv4.

The jobs resolve successful current develop artifacts to immutable identities before execution. They check the installed SDK against the current SDK runtime and exact-wheel downstream verification. Generated feature tables use actual JUnit and cleanup outcomes. The summary rejects missing reports, mixed test revisions and different artifact candidates. It records tested provider digests and scope limits without adding manual approval steps.

Verification: pipeline 29094 passes all ten jobs at fbd83e83: 250 framework checks, 21 container cases, 20 Helm cases and 30 actual EURECOM/CoreSim cases. All provider, cluster and controller cleanup checks pass. The observed feature table confirms one common artifact candidate and retains explicit limits. GitLab reports 57 phase/cleanup results separately from the underlying application cases.

The normal SRM image keeps SDK rc4; its runtime contents and declared dependency metadata match the current rc5 build. Final 2.0.0 package integration remains a separate release handoff.

Reuses the !6 (merged)/!7 (merged)/!8 (merged) histories and the existing deploy/e2e playbooks. Default collection retains its original 23 cases. API compatibility here does not establish radio behavior, effective QoS, packet steering, automatic notifications, positioning accuracy or TI port/protocol filters. No OEG, SRM or SDK runtime implementation changes are included.