Remove redundant scaling-group rewrite at allocation time

Goal

Remove the redundant scaling-group writes performed at allocation time in the scheduler repository. Sessions and kernels already receive their scaling group (name and resource_group_id) as non-null values at enqueue (BA-6644 / BA-6648), and the allocation-time value is a pure round-trip of the session's own value (PendingSessionData -> SessionWorkload -> SessionAllocation -> UPDATE back onto the same row). target_sgroup_names re-routing is never read by the scheduler, so the rewrite is dead work and a future id/name divergence risk once readers move to resource_group_id.

Scope

- repositories/scheduler/db_source/db_source.py: remove scaling_group_name= from the session resource-assignment UPDATE (~line 2110); keep agent_ids update.
- Same file: remove scaling_group= from the kernel SCHEDULED-promotion UPDATE (~line 2096).
- Optional (same file, one line): switch the enqueue access check (~line 1563) to compare AllowedScalingGroup.id against draft.scope.resource_group_id instead of names.

Success Criteria

- [ ] Allocation update no longer writes scaling_group_name / kernel scaling_group; agent assignment fields are still written
- [ ] Session/kernel scaling group values after scheduling are identical to enqueue-time values (existing scheduler tests pass unchanged)
- [ ] pants test passes for affected packages

JIRA Issue: BA-6706

Start in repositories/scheduler/db_source/db_source.py around the allocation updates near lines 2096 and 2110, then review the enqueue access check near line 1563. Remove the redundant scaling-group writes while preserving agent assignment updates; verify scaling-group values remain unchanged from enqueue time and run pants test for the affected packages.

Written by the indexing model from the issue text.