Add conversion function for MKDA-format datasets

[tsalo]

Summary

The MKDA toolbox uses its own custom format for dataset files.
We may want to implement a conversion function so that we can use MKDA-style datasets in NiMARE.

Additional details

I am not sure if there's an automatic way to distinguish metadata and annotation columns.

[tsalo]

I think the solution should be to include metadata_columns and annotation_columns parameters in the conversion function. We should also automatically convert N/Subjects to sample_sizes.