Reconcile OpenSearch test-doc stack assumptions (ES/Kibana vs real stack) #37292

Related to #37126 (where this was originally scoped and partially fixed) and PR #37258.

Description

docs/backend/OPENSEARCH_MIGRATION_TEST_PLAN.md and docs/backend/OPENSEARCH_MIGRATION_TESTER_GUIDE.md disagree with each other on what the default OpenSearch migration test environment actually is, and TEST_PLAN.md's assumed baseline stack doesn't exist anywhere in the repo.

Already fixed, no action needed: TEST_PLAN.md's "limited-user stack" section (3 of an originally-flagged 7 path occurrences) — it referenced a docker-compose.limited-user.yml file that doesn't exist; turned out every credential/port/script detail in that section already matched the real single-node-os-migration/docker-compose.yml exactly, so it was a wrong filename, not missing infra. Fixed in PR #37258, along with an unrelated service-name typo (opensearch-provision → real opensearch3-provision) found in that same section.

What's left — the actual inconsistency, still unresolved:

• TEST_PLAN.md assumes its baseline/default stack is real Elasticsearch 7.10 + Kibana ("old engine") migrating to OpenSearch 3.x ("new engine") — referenced throughout as living at docker/docker-compose-examples/os-migration/.
• TESTER_GUIDE.md treats OpenSearch 1.x → OpenSearch 3.x (docker/docker-compose-examples/single-node-os-migration/) as the primary, self-contained default stack, and describes the ES+Kibana stack as an optional second stack "if you specifically need real Elasticsearch" — one that, per TESTER_GUIDE.md itself, "does not include dotCMS or the database."
• docker/docker-compose-examples/os-migration/ does not exist anywhere in the repo. Checked git history: the migration stack has been OpenSearch 1.x → 3.x since its very first commit ((docs) OpenSearch 1.x to 3.x migration docker-compose stack #35864, "OpenSearch 1.x to 3.x migration docker-compose stack"). There's no evidence a real ES+Kibana stack was ever built.
• "ES 7.10" / "Kibana" appear ~20+ times in TEST_PLAN.md alone.

This was raised in the team's Slack channel for a decision (as of this ticket, still awaiting a reply): is the ES+Kibana stack still planned and just not built yet, or was it never meant to be separate infra — in which case TEST_PLAN.md should be rewritten around the real OpenSearch-only stack, matching what TESTER_GUIDE.md already does.

Acceptance Criteria

• Get a decision from the team on Slack: build the missing docker/docker-compose-examples/os-migration/ ES+Kibana stack, or rewrite TEST_PLAN.md to use single-node-os-migration/ (OpenSearch 1.x) as its baseline like TESTER_GUIDE.md does
• If rewriting to OpenSearch-only: update the ~20+ "ES 7.10"/"Kibana" mentions in TEST_PLAN.md, including the "Environment" table (old engine (ES 7.10), Kibana @ :5601) and every path reference to docker/docker-compose-examples/os-migration/docker-compose.yml (lines 23, 124, 250 — the remaining 3 of the original 7 flagged path occurrences, now that PR fix(docs): correct contradictions and fabricated APIs in backend AI-context docs #37258 fixed the other 4)
• If rewriting to OpenSearch-only: rework the Kibana-specific steps in the following test cases so they use OpenSearch Dashboards instead: TC-001, TC-003, TC-012, TC-013, TC-019, TC-020, TC-022, TC-023, TC-025, TC-046
• If rewriting to OpenSearch-only: update TESTER_GUIDE.md's "Elasticsearch variant" callout (lines 197-201) to stop pointing at a stack that won't exist, or remove it
• If building the real ES+Kibana stack instead: create the actual docker/docker-compose-examples/os-migration/ compose stack matching what TEST_PLAN.md already describes (ES 7.10 + Kibana on :9200/:5601, OpenSearch 3.x + Dashboards on :9201/:5602), then verify every existing TEST_PLAN.md claim against it
• Either way: make TEST_PLAN.md and TESTER_GUIDE.md agree on which stack is the default, so a reader doesn't get two different answers depending on which doc they open first

Priority

Medium

Additional Context

Full background and the original audit trail: #37126 (see its two M0-linked comments in #37125 for citations), and PR #37258 (the already-merged/fixed portion).