URL.click('?') retains the base query instead of clearing it #207

On main commit 978f2e645578aad856fe5e03e2e2c017c8c0f7ca, resolving a query-only reference with an empty query leaves the base query in place.

from hyperlink import URL
from urllib.parse import urljoin

base = "http://a/b/c?x=1"
print(URL.from_text(base).click("?").to_text())
print(urljoin(base, "?"))
print(URL.from_text(base).click("?y=2").to_text())

Observed:

http://a/b/c?x=1
http://a/b/c?
http://a/b/c?y=2

The last call is a nonempty-query control. RFC 3986 section 5.2.2 says a defined query on a reference with an empty path replaces the base query, even when the query is empty. The expected target is therefore http://a/b/c?.

URL.from_text("http://a/b/c?").to_text() also drops the trailing question mark, so the parser appears to lose the distinction between an absent query and a present-but-empty query. This affects links intended to clear existing query parameters.