Task: Delete the ores.cli component and its round-trip chain

Delete projects/ores.cli/, the domain round-trip dataset and checker, the CLI recipes, and every hard-gate reference, so the tree configures and builds without ores.cli.

• projects/ores.cli/, doc/recipes/cli/, assets/test_data/domain_roundtrip/, scripts/ore_domain_roundtrip_check.py, and the ore_coverage workflow are gone.
• A reference sweep over live paths returns no ores.cli hit.
• The tree configures and builds, and compass lint passes with no dangling id link.

Four units, each landed as its own commit and checked before the next.

1. Delete projects/ores.cli/ and the ores.cli line in projects/CMakeLists.txt.
2. Delete the round-trip chain: assets/test_data/domain_roundtrip/, scripts/ore_domain_roundtrip_check.py, and .github/workflows/ore_coverage.yml.
3. Delete doc/recipes/cli/ and fix every live reference: the dangling id links flagged by compass lint, the architecture pages, the manual chapters, the recipe inventories, and the backlog story that planned ores.cli iam tenants list.
4. Verify: a reference sweep over live paths, a configure and build, and compass lint.

Fallback: the component is one commit, so if the build finds an unexpected dependent, git revert of that commit restores it.

Manual QA scenarios (scaffolded via compass add test_scenario) that verify this task. Link new ones here as they're created; the scenario doc itself links back via its "Verifies task" field.

PR [#2140](https://github.com/OreStudio/OreStudio/pull/2140) merged on 2026-09-23 as 388a90445f. It was rebased onto ddced912a5 first, which resolved two conflicts with the Qt-remnant cleanup, in doc/meta/tag_inventory.org and projects/ores.hpp. Two independent verifiers returned PASS+NOTES, one before the rebase and one after. All six PR checks passed.

Done on 2026-09-23, branch feature/decommission-ores-cli, off origin/main at 8d2565db07.

Deleted, in three commits:

• projects/ores.cli/, 129 files and 12,922 lines of C++, plus its add_subdirectory line. Nothing linked ores.cli.lib, and no codegen model, template or drift registry named it.
• The round-trip chain: assets/test_data/domain_roundtrip/ (1474 XML files), scripts/ore_domain_roundtrip_check.py, and .github/workflows/ore_coverage.yml.
• doc/recipes/cli/, 65 files, and every live reference to them.

Checked against the acceptance criteria:

• Deleted. All five are gone.
• Reference sweep. git grep over live paths returns no hit outside historical sprint archives, dated plans, discarded backlog entries and the kanban test fixtures. Those keep their text as the record, and the backlog items that planned CLI work now carry a retirement note and point at ores.shell.
• Build. cmake --preset linux-clang-debug-make configures and cmake --build --preset linux-clang-debug-make -j 8 exits 0. The build was a full rebuild of 3,739 objects. ores.shell.exe and the service executables link and build.
• Lint. compass lint exits 0 with no dangling id link. The 158 durable-to-agile links are a pre-existing advisory, unchanged.

Not run: ctest. The environment is stale (.env is version 20 against a required 22, and 23 services are missing), so the database-backed suites cannot run here. The build and the sweep are the evidence for this change; the deletion removes code rather than altering behaviour.

Follow-up recorded, not done here: ores::ore::xml::exporter::roundtrip lost its only caller, and ores.service::run_host_sync now has one. Both are tracked in the story's out-of-scope list.