My take on this is that if your "usage" info is more than fits on a single screen, then it is too much and should be documented separately.

Usage text should be an on-screen prompt of the syntax of the command that the user can put on the screen to remind him while he is typing the command.

If the text scrolls off the top of his screen, he can't do that. In which case he's better off loading the help info--man; info; html; whatever--in a separate screen from where he is typing the command, so that he can refer between them.

My real bottom line is that if the command syntax is so complicated that it cannot be outlined in less than one screenful of text, it would be much better to allow the user to supply the command parameters and switches through a file. Eg: command -p job.parms

And perhaps supply a separate interactive utility or gui to construct those parameters files.