Create Transformer to combine coordinates from different experiments in the same studies

[tsalo]

Summary

Per Turkeltaub et al. (2012), treating separate experiments from the same subject group as independent introduces bias. We haven't really tackled this problem in NiMARE, because we don't have any identifier for individual subject groups. However, what we do have are study and experiment IDs. Different experiments from the same study can easily be (1) from different subject groups or (2) from incompatible contrasts, so I don't think it's a good idea to automatically merge these coordinates within, say, the ALE Estimator.

What I am proposing is a new Transformer, that would identify any studies with multiple experiments in a Dataset, and merge them into a single experiment (if they have matching metadata, such as sample size). Users would then have to elect to do this prior to their meta-analysis.

Additional details

This would reduce the influence of studies with multiple experiments on coordinate-based meta-analyses.

Next steps

1. Create a new Transformer, perhaps in nimare.diagnostics.
2. Identify any crucial metadata for differentiating subject groups in a study. The only thing that comes to mind at the moment is sample size, but there may be other relevant info.
3. Add information about the Transformer to the documentation (e.g., an example).

Open questions

• I'm not sure what to do with other fields, like texts, annotations, and images.

[tsalo]

One minor thing- I believe that Sleuth automatically combines experiments within studies, so Sleuth users wouldn't have to worry about this. It would mostly impact folks using NIMADS-based Datasets.