[Documentation] Suggestion for improvement

[kaitj]

Not sure which repo this issue should go under, but this one seems to probably maybe fit the bill.

The current Styxbook serves more as a "Getting started to Niwrap" with instructions on setting up the runners and short (toy) examples. For actually using Niwrap, I find myself still diving more into the repository or relying on the autocompletion for using the tool itself. This isn't as big of a deal for the simple commands (e.g. bet), but is more cumbersome for those with any type of subcommands (e.g. ants, workbench, greedy, etc).

Talking with internal users, one suggestion was to improve niwrap documentation associated with this. Pulling this as an example from one of my scripts:

resampled = ants.ants_apply_transforms(
        dimensionality=3,
        input_image_type=3,
        input_image=in_img_fpath,
        reference_image=ref_img_fpath,
        interpolation=ants.ants_apply_transforms_nearest_neighbor_params(),
        output=ants.ants_apply_transforms_warped_output_params(output_img)
    )

It's not obvious, for example, that in ants, the output should be ants.ants_apply_transforms_warped_output_params or that interpolation should pass these different param methods (as shown above and also differ whether or not actual params are being passed to it or not). Compared to bet, where these values can be passed directly. As of v0.6.x, the autocomplete also shows the associated class iirc (e.g. ants.AntsApplyTransformsNearestNeighbor - not sure this is a real one, but just an example).

All that said, I'm not sure how that could be clearly documented given the sheer number of commands, but I think some way of making this more transparent or easier to find could help this easier to use.