Don't allow "UTF-16 surrogate codepoints" in char or str

[jruderman]

I think

> "\uD800"
"\ud800"

should be disallowed at compile time and

> 0xD800 as char
'\ud800'

should assert. Or even better, the incomplete int as char could be removed.

[jruderman]

At least it's consistent now:

> let mut a = ~"hi"; a.push_char(0xD800 as char); a
~"hi\ud800"

[jruderman]

Because UTF-8 encodings of these surrogates are not considered "valid UTF-8" https://en.wikipedia.org/wiki/UTF-8#Invalid_code_points

Because OS APIs often do weird/unsafe things if you give them mismatched surrogates, or surrogates where they aren't expecting surrogates.

[thestinger]

Fixed by 62a3434 (char) and b153219 (str). Please open another issue if there are still any holes (the string index-assign one is reported).

[jruderman]

The cases in this bug report are both fixed.

> \uD800
error: illegal numeric character escape

and

> (0xD800 as char)
error: only `u8` can be cast as `char`, not `<VI0>`

Thanks, @thestinger!