Text on reordering and Objects going from DNE to Exists is incorrect

[ianswett]

While writing #1342, it was observed that the following text doesn't seem correct anymore.

"Note that due to reordering, an implementation can receive an Object after
receiving an indication that the Object in question does not exist. The
endpoint SHOULD NOT cache or forward the object in this case."

This text might not be needed?

[sharmafb]

Why is this section of text incorrect?

[afrind]

There are cases where a publisher is clearly breaking the object model, that is malformed, e.g.:

10,0, 10,1, 10,2, END_OF_GROUP, 10,4 <-- This is malformed

But here's a scenario that seems legit:

S->P: SUBSCRIBE
P->S: SUB_OK
Publisher creates object 0,0, sends it on a subgroup, then "forgets" it.
Subgroup delayed in network
S->P: FETCH 0,0
P->S: FETCH_OK, Empty Response -> 0,0 does not exist
0,0 arrives at S -> this is not malformed.

Our text is currently says both things and we need to clear this up.