HPC queue cleanup not working correctly

#### Problem Description

We had a job with a large number of tasks added some time ago. This caused the job cleanup to fail and the queue to build up.

We debugged the issue in the underlying database. All queries in HpcScheduler.dbo.sp_DeleteOldJobs stored procedure are executed as expected up to "delete from TaskGroup..." which takes a very long time to execute. Our DBA tracked it back to a constraint on one of the tables that was looking up TaskGroupDependency. There is a missing index on the TaskGroupDependency table - see below.

USE [HPCScheduler]
GO

/****** Object: Index [IX_TaskGroupDependency] Script Date: 20/09/2025 13:33:23 ******/
CREATE NONCLUSTERED INDEX [IX_TaskGroupDependency] ON [dbo].[TaskGroupDependency]
(
[ParentId] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON, OPTIMIZE_FOR_SEQUENTIAL_KEY = OFF) ON [PRIMARY]
GO

This change also significantly improves the performance when deleting all jobs from the queue.

#### Steps to Reproduce

Add job with large number of tasks with a lot of interdependency.

#### Expected Results

Job is removed when queue cleanup starts

#### Actual Results

Job is not removed (nor are any other jobs in HPC Pack 2016)