repr() of a relative URL does not round-trip: eval(repr(u)) != u #210

For a URL without a scheme, reading back its printed form gives an unequal URL:

>>> from hyperlink import URL
>>> u = URL.from_text("a b")
>>> u
URL.from_text('a%20b')
>>> URL.from_text("a%20b") == u
False

With a scheme it works: URL.from_text("http://example.com/a b") prints back unchanged and compares equal.

Cause: in _encode_path_parts, the first segment of a scheme-less path is encoded without passing maximal, so to_text() encodes it fully.