Review results aggregations for PBM

[yiwen-h]

Review aggregations in OP for PBM - do we need the same levels of aggregation in results for PBM as we do for provider level? We could for example only provide age_group not age in PBM results, or tretspef_grouped not tretspef

Current aggregations mean that OP data is too big to run in PBM - we're having to make compromises e.g. with The-Strategy-Unit/nhp_model#473

[yiwen-h]

Problem with having different aggregations is that outputs app and related projects using results may struggle. I think we should keep results consistent to avoid spaghetti situations