Bug - handling last sentence in task1_converter.py

[antonyscerri]

Hi

The conversion script task1_converter.py does not handle the last line in all the deft_files where they dont end in a blank line (which is all those in the train subdirectory). The code isn't checking for a new sentences concatenation after going through all the lines.

This brings up another question which is the corpus size in terms of sentences. I've not been able to match up with the figures in the paper against any of the sets of files in this repo, so i wanted to check how many sentences should there in fact be in total.

Thanks

Tony