Split diffcalc workflow to add concurrency group

[smoogipoo]

The diffcalc runner currently dies when many runs are queued up because they're running in lock-step and each one eventually manages to extract scores/beatmaps prior to other runs having completed.

The plan of attack here is to split the workflow into two pieces: the main diffcalc.yml and a new reusable _diffcalc_processor.yml workflow with concurrency applied that effectively locks for the duration of the workflow on the self-hosted runner.

I've tried to make as minimal changes as possible for now. Most of the changes are around the "environment" job because there's a level of indirection for data like the PR URL, comment text, and dispatch inputs.

Here's a dispatched run: https://github.com/ppy/osu/actions/runs/11756297523

Keep in mind I haven't tested the PR/comment part of this though I've made a good effort to ensure it should work. I'm hoping this can be first merged with minimal friction and I can commit directly to master any follow-up fixes that need to be done (it's hard to test that part out in full...).

[smoogipoo]

I've done this because gha doesn't have any way to reorder/organise workflows, and this at least pushes it to the bottom of the list: https://github.com/smoogipoo/osu/actions/workflows/diffcalc.yml