Allow message specification for should_fail

[sfackler]

The test harness will make sure that the panic message contains the
specified string. This is useful to help make #[should_fail] tests a
bit less brittle by decreasing the chance that the test isn't
"accidentally" passing due to a panic occurring earlier than expected.
The behavior is in some ways similar to JUnit's expected feature:
@Test(expected=NullPointerException.class).

Without the message assertion, this test would pass even though it's not
actually reaching the intended part of the code:

#[test]
#[should_fail(message = "out of bounds")]
fn test_oob_array_access() {
    let idx: uint = from_str("13o").unwrap(); // oops, this will panic
    [1i32, 2, 3][idx];
}

[sfackler]

cc @steveklabnik - we talked about this on IRC yesterday.

[alexcrichton]

Interesting! This was somewhat surprising to me as I thought this would be a custom message to print out on test failure rather than "this test failed" with no output from the test. The semantic meaning implemented here, however, I think makes more sense to me.

This is one of the major reasons I avoid should_fail tests as much as possible because I'm not actually sure that the part I wanted to fail actually failed, so this seems like a great thing to have. @sfackler do you think that my initial interpretation is a possible to make, or am I off my rocker? Perhaps we could get "expected" somewhere in the name?

[sfackler]

Sure, changing message to expected seems reasonable.

[sfackler]

@alexcrichton updated. Also added a bit to the testing docs.