Fix for concurrent modification exception on removing table

[podile]

Fix for the issue #77

I am able reproduce the issue with below code snippet.

`package com.zegelin.cassandra.exporter;

import com.google.common.collect.Sets;
import java.util.Set;

public class TestSets {

public static void main(String args[]) {
    Set<String> set1 = Sets.newHashSet("E1", "E2", "E3", "E4", "E5", "E6");
    Set<String> set2 = Sets.newHashSet("E4", "E5", "E6");
    Set<String> removedMBeans = Sets.difference(set1, set2);
    for (String e : removedMBeans) {
        set1.remove(e);
    }
}

}
`

[podile]

This PR fixes the issue #77

[eperott]

Would it be possible to reproduce the issue in a unit test? That way it would be easy to verify the fix, and to make sure that the issue is not re-introduced in a future refactoring.

Also, your PR contain a range of different changes. Perhaps it will be easier for maintainers to review/merge your PR if you can extract your fix into a separate branch before you create the PR.

[podile]

Would it be possible to reproduce the issue in a unit test? That way it would be easy to verify the fix, and to make sure that the issue is not re-introduced in a future refactoring.

Also, your PR contain a range of different changes. Perhaps it will be easier for maintainers to review/merge your PR if you can extract your fix into a separate branch before you create the PR.

Lot of mocking required to initialize and call the method reconcileMBeans of the class JMXHarvester as the constructor itself triggering lot of functionality through metadataFactory. Since the fix I made is very simple and straight forward,I would prefer not to add any test case in this particular case.

Sorry for the inconvenience made by the rage of changes in my PR, I pulled all merged changes form zegelin/cassandra-exporter as it contains important fixes. I will find a better way next time.