Clover fails to find coverage snapshot files when IBM JRE is used

Problem: it seems that IBM JRE implementation differs from Oracle JRE in such way that Object.hashCode() can return a negative value. As a consequence, file name suffix generated for coverage snapshot and test slice snapshot may be composed not only of '0-9a-z_' characters, but also have minus "-" sign in it.

As a result such file name does not match regular expression used for searching of coverage files and coverage is not read.

Workaround: rename coverage snapshot file(s) and remove "-" character.