Reimplement test data in form of `evaluationContext: expectedEvaluationResult`

[khvn26]

Additionally, think on any of the following:

• Better naming for expected feature values "foo" -> "overridden_by_segment_1". This might be low priority as we'll have the "reason" field to check.
• Using JSONC to document and explain the test cases

[khvn26]

As part of this, we should make test data conform with Flagsmith/flagsmith#6124 (done in 45cc7eb + c9343de).

[khvn26]

As part of this issue, we'll break the humongous file to individual test cases.