ID for language-invariant strings

[Frank Ellermann]

Doug Ewell wrote:

> anything other than:
> * a new ISO 639 code element
> * 'invariant'
> * 'zxx'
> * 'und'
> would be flat wrong, and should not be discussed further.

For almost-English 'i-default' might work, '' (empty)
can be better than 'zxx' where empty tags are allowed,
to discourage overzealous translators 'en-x-fontname',
or 'x-fontname' might do the trick.  The 'invariant' 
is too long, a hypothetical 'invar' would be limited
to BCP 47 applications, same issue as for 'i-default'.  

 Frank