[Checkers] Inconsistent numbers in Figure 4 of Checker Framework paper

[Michael Ernst]

Matt-

> The reason why the NNEL sig numbers are higher [than previously] is that the
> old checkers failed to warn about invalid overriding in some places,
> so the new checkers (correctly) required me to add some annotations to
> fix bad overrides.
> 
> The solution:
> If we had redone the NonNull case studies with the new checkers, we'd
> get the exact same number of signature annotations as we got for NNEL.
> Therefore, I'd suggest copying the "sig" column for NNEL to NonNull
> and then adjusting the NonNull "total" column accordingly.

OK, I have made this fix.  Thanks a lot, and enjoy the rest of your
vacation!  (I really hope we won't need any more answers from you...)

                    -Mike