test runner tasks could take their name from the test's name

E.g.

``` rust
#[test]
fn test_foo() {
    assert_eq!(1, 2);
}
```

would be more useful it if it printed `task 'test_foo' failed at 'assertion failed: ...` rather than `task '<unnamed>' failed at 'assertion failed: ...`.

This would make finding the cause of a `FAILED` in the spew of `#[should_fail]` `fail!`s easier, at least until we get capturing/redisplay-as-necessary of the output.
