Skip inconsistent test (nltk#2605)

* Skip inconsistent test

There is a ~1.8% chance that this test will randomly fail, despite working correctly.

* Reduced precision of tagger evaluation test

Rather than skipping, reducing the precision to between 70% and 80%, rather than between 73% and 74% seems like the best solution for preserving the test, while keeping the test easily readable.

nltk/tag/__init__.py

@@ -59,7 +59,7 @@
 We evaluate a tagger on data that was not seen during training:
 
     >>> tagger.evaluate(brown.tagged_sents(categories='news')[500:600])
-    0.73...
+    0.7...
 
 For more information, please consult chapter 5 of the NLTK Book.
 """