Task: Retire the JSON-era residue the org migration left behind

Retire the residue the JSON-to-org migration left behind: delete the dead migration scripts, correct the stale skill and template guidance, and clear the dead references to the deleted models tree.

• No script in the tree points at projects/ores.codegen/models/.
• No skill source, recipe, or template org source tells the reader to author a JSON entity model.
• compass build --direct skills and compass build --direct tangle_codegen_templates both run clean.

1. Delete the five spent migrate_*_to_org.py converters.
2. Repoint the three codegen skills and the three recipes at the org model.
3. Fix the template org sources, then re-tangle.
4. Apply the same text by hand to the three committed generated populate SQL files. Their dataset pipeline is no longer runnable, so the template is the only source of truth.

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

Five converters deleted. Nothing in the tree imports them.

Repointed at the org model: compass-codegen-add-entity, compass-codegen-add-sql-schema, compass-codegen-add-qt-entity, how_do_i_create_a_new_entity, how_do_i_create_entity_schema, how_do_i_add_a_new_codegen_entity_model, and two analysis references. The --model help text in cli.py no longer shows the deleted layout.

Templates re-tangled. The four .mustache diffs are exactly the intended lines and nothing else. Three generated populate SQL files carry the same change by hand, and the slovaris methodology.txt payload now names its own directory.

Left alone on purpose: the four uncalled generate_*.sh scripts still name generator.py. They call a file that does not exist. The decommission_codegen_bash_scripts capture in the backlog already owns them, so they are not this task's work.

Abandoned at sprint 25 close. Sprint 26 replaces the codegen approaches tried so far with one method: for each module, sync codegen and clear its drift. Sprint 26 opens new, clean stories for each module. This story is not carried forward.