clarity needed for the non-sealed keyword

[Florian Weimer]

* Remi Forax:

> a sealed type lists all it's subtypes (the permits clause) so the
> name of a subtype have to be stable, i.e. not change from one
> compilation to an other or from one compiler to an other.  So you
> can not use a local class as a subtype of a sealed type because the
> name of a local class is not stable.
>
> So there is no ambiguity for the tokens "non-sealed" inside a
> method, it always means "non" "-" "sealed".

But why does the keyword contain a “-”?  Why wasn't “nonsealed” good
enough?  Is it really worth creating this special case for the lexer?