Drop scaling_group_name column from session row

Goal

Remove sessions.scaling_group_name now that all writers and readers use resource_group_id (or resolve the name via join). Final step of the session-side id migration; kernels/agents legacy name columns and the PK swap are handled in BA-6050.

Scope

- ORM: remove SessionRow.scaling_group_name; rewire the scaling_group relationship to the resource_group_id FK (models/session/row.py ~:456-461).
- Alembic: drop the column. Optional: also drop the dead target_sgroup_names column (list of names, never read by the scheduler) to clear name remnants in one pass.
- Sweep remaining compile-time references; SessionData.scaling_group_name stays as an API-compat field populated via the join/relationship.

Success Criteria

- [ ] sessions table has no scaling_group_name; sessions reference scaling groups by resource_group_id only
- [ ] Full session lifecycle works on the live server: create -> schedule -> app launch -> terminate
- [ ] All GQL/REST session list/filter/order queries pass
- [ ] pants test passes for affected packages

JIRA Issue: BA-6713

Start in models/session/row.py around lines 456-461, then locate the Alembic migration and sweep remaining compile-time references to SessionRow.scaling_group_name. Verify that SessionData.scaling_group_name remains API-compatible, the sessions table uses resource_group_id only, and the full lifecycle, GQL/REST queries, and affected-package pants tests pass.

Written by the indexing model from the issue text.