Fix index slice in ExtensionPointChangedEvent when plugin changes

[kitchoi]

When the extensions list is changed by adding or removing a plugin, the ExtensionPointChangedEvent reports a slice object as the index, but the end value is incorrect: If one uses that index to reproduce the new list from the values of the old list, they get incorrect results.

The first commit adds two assertions. The first assertion fails with master like this:

Traceback (most recent call last):
  File "/Users/kchoi/Work/ETS/envisage/envisage/tests/test_extension_point_changed.py", line 268, in test_assign_non_empty_list
    listener.new.removed, source_values[listener.new.index]
AssertionError: [1, 2, 3] != [1, 2, 3, 98]

That is, if you use the index on the old values to recover the removed values, you get the wrong answer.

The second assertion fails with master like this:

Traceback (most recent call last):
  File "/Users/kchoi/Work/ETS/envisage/envisage/tests/test_extension_point_changed.py", line 274, in test_assign_non_empty_list
    self.assertEqual(source_values, application.get_extensions("a.x"))
AssertionError: Lists differ: [2, 4, 6, 8, 99, 100] != [2, 4, 6, 8, 98, 99, 100]

First differing element 4:
99
98

Second list contains 1 additional elements.
First extra element 6:
100

- [2, 4, 6, 8, 99, 100]
+ [2, 4, 6, 8, 98, 99, 100]
?             ++++

That is if you use the index on the old values and assign the "added" values to it, you also don't recover the new values.

The second commit fixes both assertions, and make the necessary change to the test that asserts the slice end value.

[aaronayres35]

Do we want to try to push this through for the 5.0.0 release, or is it intended to wait like #354?

[kitchoi]

I don't think this needs to wait like #354. @rahulporuri what do you think?

[kitchoi]

On the other hand, if this bug has lived this long (13 years) and no one realized it, it is probably not important enough for it to block the upcoming release.

[rahulporuri]

LGTM. Let's push this through in this release.

[rahulporuri]

@kitchoi sorry for the enormous delay on this PR.

[kitchoi]

No problem at all. Thank you!

[kitchoi]

I believe this has not gone into the maintenance branch, so I am deploying the 'need backport to 5.0' tag.