[tei-council] Minor changes in ND

[Lou Burnard]

Most of this afternoon was spent working on the model classes in ND, I fear.

offset and geogFeat only make sense *within* a placeNamePart: they are 
not placeStateLike in themselves. 
("<location><geogFeat>river</geogFeat></location>"?  i think not)

I have now defined a new phrase-level class model.offsetLike, to which 
they belong and removed them from placeNamePart and (in the case of 
geogFeat) place.stateLike. I also gave them both a content model of 
macro.xtext.

I added  geogName to placeNamePart (I think geogFeat was there in error 
for it)

I made placeNamePart a subclass of placeStateLike and eliminated the 
duplication of class memberships for all its members.

I added some prose to the chapter about the various classes, and removed 
some lies about what's permitted where. (Basically, now almost anything 
is permitted almost anywhere)

Other minor tweaks:
- Is <population> a state or a trait of a place? The prose says it's a 
state, the schema that it's a trait. It's less work to change the 
former, so that's what I did.
- @level of langKnown was defined as data.code which is a pointer; since 
there's no information about what on earth it's supposed to point to 
(and several examples used a bare word anyway) I changed it to data.word
- removed  @scheme from location - a historical hangover

This chapter still needs some examples, and the line of argument is 
occasionally not as clear as it might be. But it is still a wonderful 
chapter and there are lot of others...