It's a feature. Hit.new_from_read_builtin reads one "hit", which in most file formats corresponds to a single line. Bunch.new_dict_from_read_builtin reads in an entire bunch or set of bunches, so reads in the entire file. I could imagine some feature in Hit.read_blah to auto-detect when the read is on a header line and ignore it. But the difference is Hit read function is reading line-by-line whereas bunch read function is reading file-by-file.

Does that answer the question?

Thanks.
I realized that Hit.read works line by line but was not sure if it was designed to return an error
on encountering a header line -- instead of maybe skipping or returning something which can interpreted
as a header read; but that's probably more complicated than code just skipping before calling Hit.read