Improve JavacBindingResolver

[snjeza]

Related to #755

[snjeza]

de42c03#diff-af030ac91c9fb35cb660aa9076bd04fcbfde2ebc73c583664b39293fc8475e3bR449-R455 doubles the speed of the resolve method.

[datho7561]

There are around 13 new failures, I think these are them:

org.eclipse.jdt.core.tests.dom.ASTConverter18Test.testBug433879c
org.eclipse.jdt.core.tests.dom.ASTConverter18Test.testBug433879d
org.eclipse.jdt.core.tests.dom.ASTConverterAST3Test.test0261
org.eclipse.jdt.core.tests.dom.ASTConverterAST4Test.test0261
org.eclipse.jdt.core.tests.dom.ASTConverterAST8Test.test0261
org.eclipse.jdt.core.tests.dom.ASTConverterTest2.test0438
org.eclipse.jdt.core.tests.dom.ASTConverterTestAST3_2.test0438
org.eclipse.jdt.core.tests.dom.ASTConverterTestAST3_2.test0667_2
org.eclipse.jdt.core.tests.dom.ASTConverterTestAST4_2.test0438
org.eclipse.jdt.core.tests.dom.ASTConverterTestAST4_2.test0667_2
org.eclipse.jdt.core.tests.dom.ASTConverterTestAST8_2.test0438
org.eclipse.jdt.core.tests.dom.ASTConverterTestAST8_2.test0667_2
org.eclipse.jdt.core.tests.dom.ASTConverterTest.test0261

[datho7561]

These two tests are invalid argument exceptions thrown by javac:

org.eclipse.jdt.core.tests.dom.ASTConverter18Test.testBug433879c
org.eclipse.jdt.core.tests.dom.ASTConverter18Test.testBug433879d

I think they are worth investigating further, because perhaps we are passing in bad values to javac.

The other failures appear to be changes in the number of errors reported by javac. I am less worried about those ones.

[datho7561]

Seems good to me. Thanks, Snjezana!

[mickaelistria]

The other failures appear to be changes in the number of errors reported by javac. I am less worried about those ones.

I was worried but debugging shows that now with this change, the UnusedTreeScanner is run although there are errors in enter instead of being interrupted earlier. I think it's a good thing. If it becomes bogus, we'll have to add a check before running .analyze(tree), but so far so good. Thanks!