Fix evaluation of unicode character arrays (char16_t[] and char32_t[])

Suppose we have the UTF-16 string:
    char16_t[] s = u"hello";
Before this patch, evaluating the string in lldb would get:
    (char16_t [6]) $0 = ([0] = U+0068 u'h', [1] = U+0065 u'e', [2] = U+006c u'l', [3] = U+006c u'l', [4] = U+006f u'o', [5] = U+0000 u'\0')
After applying the patch, we now get:
    (char16_t [6]) $0 = u"hello"

Patch from evgeny.leviant@gmail.com
Reviewed by: granata.enrico
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13053

llvm-svn: 248555