Manually coded SVG files are short and their code can easily be edited using a simple text editor.
Two diverging opinions are widespread
1st: The SVG code should be formatted mainly that a human person can read it easily, no matter how large the file would grow. Excessive use of SVG elements as line, rect, circle, ellipse, polyline, polygon instead of path represents best the geometric structures. Lots of spaces, tabs and linefeeds give a good appearance, together with sufficient comments it will help to a swift overview.
2nd: The coding that is passed to the librsvg should consist only of the executeable statements. Since manual SVG graphics are rather restricted to simpler structures it is less necessary to explain a lot; any comment and edit help may be in the file description page, or the file talk page even excessively, but the file itself should kept puristicly small in a compressed format.