A purely random number is almost certainly always beyond copyright, but as soon as someone puts limitations on that randomness a court may find that enough.
And what matters for a takedown is not the number, but the actually document being published. Github is not hosting the random number. It is hosting that number in the context of a larger document and it is that document that is subject to the takedown request. Things might be different if github hosted only the number without the associated labels and code.