Added two keyword arguments to Template.__init__(): "default_encoding" and
"decode_errors".  These are passed internally to unicode() as the "encoding"
and "errors" arguments when converting strings of type str to unicode.

From the issue:

    The constructor `Template.__init__()` should accept an optional
    `escape` function argument instead of the boolean `disable_escape`.
    This is more natural.

    Examples:

     * Users wanting to disable escaping can pass `lambda s: s`.
     * Non-markupsafe users wanting double-quotes escaped can pass
       `lambda s: cgi.escape(s, True)`.  The cgi.escape() function
       doesn't escape double-quotes by default.
     * Similarly, markupsafe users can specify an alternative to
       `markupsafe.escape()`.

Added a passing test case to cover issue #53.