Followup to commit 97dfa0d5 which
had the side effect to break some attribute unit tests

Add some more tests to make sure that pulling in data- properties on an element with no data set, works.

Make sure that .data() (no args) returns a list of all the data- properties as well. Also make sure that accessing a data- property via .data() doesn't cause it to change again at a later time (it should be static). Fixes #7222, #7223.

We removed deleteExpando after 1.4.2, for some reason. This caused problems with removeData() (no arguments). Fixes #7209.

Handle the case where JSLint complains about arguments in try/catch already being defined (we use the name 'e' consistently for catch(e) - will work to standardize on that now).

Bringing back the change from 80a4178a, adapted to handle both the function collision and the issue mentioned in #7196.

Getting weird results for font size tests across platforms - easier to test something else (the exact property isn't important for this test).

Forgot to update the Rakefile and build.xml files from the previous commit. Thanks to @jitter for the catch.

Make sure that we don't attempt to handle scrolling when the node is disconnected from the document. Fixes #7190.

Change font size to 15px in the test to fix a false negative. Thanks @jitter for the catch in 73e4ef27.

beforedeactivate will fire before focusout - use that to our advantage when working with the change event.

We want to make sure that we only run the MSIE-specific filter tests if opacity is explicitly not supported.

Fix some gaps in the handling of event data properties. Thanks to @jitter in ff6ceadb for the catches.

Fix patch for #7150, which was not deleting the events object properly on plain JS objects. Thanks to jitter for catching it.

The child selector-prefix branch doesn't work with selectors that have a larger depth. Removing for now - will need to re-evaluate in the future.

The world isn't ready for moving the Ajax methods to jQuery.ajax.*. Hope to move them there some day. Fixes #7146.