[developers] token-level ambiguity in the LKB parser

[Ann Copestake]

Just as a follow-up - it would be good to have some artificial test cases for 
input token-level ambiguity which exercised all the possible cases.  The 
potentially problematic issue for the LKB is the interaction with multi word 
entries (in the sense of entries with an orth value that can cover more than 
one of the initial tokens).

Ann