Refactor scoring test scene for ruleset extensibility (and move existing instance to osu! ruleset project)

[bdach]

Continuation of preparation for #23763

To work with the classic scoring algorithm, I want to get all rulesets under test coverage similar to the one we already had in TestSceneScoring. As it turns out after some checking I did on Wednesday, rulesets do wildly diverge in scoring behaviour, so I do believe that this wasn't a fruitless endeavour.

To prepare for the above, this PR refactors a base class for the scoring tests that shall live in osu.Game, and moves the existing instance from osu.Game.Tests to osu.Game.Rulesets.Osu.Tests (since it was really ever testing osu!-the-ruleset scoring anyways).

In the end I decided to implement each score calculation independently and not reuse LegacyScoreSimulators. That is mostly because the two have different goals and different levels of detail; the simulators are mostly concerned with perfect scores only, but handle all object types, while the scoring-test-local implementatons are concerned with non-perfect scoring too, but generally only handle the main object types (osu! hitcircle / taiko hit / catch fruit / mania note). That's because I don't think accurate replication there matters; classic scoring isn't going to match score V1 anyhow because of the constraints we placed on it (must be derived from standardised and not reorder scores), so the goal is to get the ballpark roughly correct.

Aside from the refactoring, this also adds some extra bells & whistles, such as being able to set particular scenarios via test steps, as well as adjust the SliderMultiplier constant on-the-fly (previously hardcoded to convenient value).

Apologies for the LoC here, but this should make it much easier to review subsequent changes.

Testing resources

osu-scoring-resources.zip contains a bunch of files, namely:

• .osu files that are beatmaps with particular test cases (all containing 100 objects)
• .osr files that correspond to replays set on beatmaps with the same name
• .txt files containing extracts of stable total score after each judgement to cross-check.

Six different test cases are provided:

• perfect score:
  • ScoreV1Perfect
  • ScoreV2Perfect
• score with misses:
  • ScoreV1MissesAt25And50
  • ScoreV2MissesAt25And50
• score with misses and 100s:
  • ScoreV1MissesAnd100s
  • ScoreV2MissesAnd100s

which correspond to the test steps added to the scene. For cross-referencing, set ScoreMultiplier = 3 in the test.

Stable code for reference:

• Score V1: https://github.com/peppy/osu-stable-reference/blob/1531237b63392e82c003c712faa028406073aa8f/osu!/GameModes/Play/Rulesets/Ruleset.cs#L1117-L1119
• Score V2: https://github.com/peppy/osu-stable-reference/blob/1531237b63392e82c003c712faa028406073aa8f/osu!/GameplayElements/Scoring/Processors/ScoreProcessorOsu.cs

[peppy]

I've checked this on surface level, seems okay. Not putting too much thought into code quality checks because test scene.