What would you like to be added:

To enable graduation of the WorkloadRequestUseMergePatch to Beta we need ensure the Workload conditions are not lost due to patches sent by scheduler:

https://github.com/kubernetes-sigs/kueue/pull/6765/files#diff-fb3d6b36d3cd727e1dd2353fb86c0aee0bfaae1c900e0f005ab913620021d551R663-R666

Introduced in the #6765

One idea is to use "strict=true" mode, but for that we need to:

1. investigate the performance impact, analyze the results from the perf tests
2. introduce some retry strategy so that the scheduler could Get the new workload (from informer cache) before patching. See related: Introduce Retry mechanism to PATCH client #6992

Why is this needed:

Overriding conditions set on workload by other controllers is very tricky.