MICRO-API for test harness

The tests are in DOMtests:

var DOMtests = [
	{
		// test 1
	},
	{
		// test 2
	},
	// etc.
];

These tests are done in the order specified here. That sometimes matters.

There are two complicated cases: user questions and asynchronous tests:
-	User questions are asked after all the tests have been done. This may matter
	if the element the question is about is changed in a subsequent test.
-	Async tests are done after a delay specified in the test. This is sometimes
	necessary (XHR, for instance), and at other times it's useful to give the
	browser some time to adjust the DOM tree.

Each individual test can have the following properties:

output				
	string
	The header under which the result is shown (generated automatically)
outputtext			
	string
	The text shown with the result. If absent it's the same as output.
test
	function				
	The actual test function. Should return a value that can be compared to an 
	expected value.
expectedValue		
	string; possibly HTML
	The value the test function is expected to return. If the function returns
	another value, the test is Wrong.
caseInsensitive
	true
	The expected value and returned value are forced to lower case and then 
	compared.
checkReturnValue
	true
	Indicates this test should return a truey value. The test succeeds if it 
	does.
shouldNotWork
	true
	Indicates that this test should give an error. It essentially reverses the
	test result: true to false and false to true
description			
	string
	A descriptive text to be shown with the test results. Has no bearing on the 
	actual test.
optional
	true
	Indicates this question does not count for the final result. Used mainly
	for vendor prefixes.
question
	object
	Defines a user question. See below.
async
	object
	Defines an async test. See below.

The following properties are generated by the test harness

result
	'worked', 'error', 'undefined', 'wrong'
	The result of the test. 
		'worked' means the test succeeded
		'wrong' means the test returned a wrong value
		'undefined' means the test returned the undefined value
		'error' means the test threw an error
returnValue
	any
	The value returned by the test function. Is usually compared to 
	expectedValue.
	
USER QUESTIONS

You can add a question object to a test, and the user will be asked a question. 
A Yes is interpreted as a succeeded test; a No as a failed test. (Using 
shouldNotWork reverses this.)

Each question object can have the following properties

questionText
	string
	The text the user sees and is asked to answer by Yes or No.
fn
	function
	A function that runs just prior to the question being asked. It should 
	receive testArea as an argument; this is the area where you can show an HTML 
	snippet to the user.
fnAfter
	function
	A function that runs after the user has answered the question. Useful for 
	cleaning up.

Note that first all main tests run, and the user questions come only afterwards, 
in order of the main test objects. This may matter if you run several tests with 
the same HTML test element.

The following property is generated by the test harness:

result
	any
	The result (returned value) of the main test. This is useful for returning 
	an object that should be shown to the user; for instance:
	test: function(id) {
		// test something
		this.question.result = document.getElementById('theTestElement');
	}
	... and then ...
	fn: function(testArea) {
		testArea.appendChild(this.result);
	}

ASYNC

Some tests require asynchronicity. To label a test asynchronous, make sure to 
set an async object on the main test object. This alerts the test harness to an 
upcoming asynchronous test.

async is an object that can contain the following properties:

ignoreMain
	true
	Indicates that the main test function should not be reported. Only 
	the asynchronous test counts here.
	If you do not set this property the main test and the asynchronous test will 
	be treated as two separate tests.
any test property
	You can repeat any property of the main test object in an async object. The 
	new value will be used for the asynchronous test; the old one will be 
	used for the main test and discarded afterwards.
	
The asynchronous test function MUST go inside a special async() function call at 
its normal place in your test script. The harness needs this function to run for 
administrative purposes.

async(test ID,test function);

For instance, this function tests the readyState property of XHR:

test: function (id) {
	var x = new XMLHttpRequest();
	x.open('POST','test.txt');
	x.onreadystatechange = function () {
		async(id,function() {
			return x.readyState;
		});
	};
	x.send();
}

Note the id arguments in the main function and async(). They're required; they make sure the asynchronous test result is properly reported.