fix: Retrieve metadata correctly from importlib_metadata (#1115)

* fixes #977: Retrieve metadata correctly from importlib_metadata

Running twine with `PYTHONWARNINGS=error`, DeprecationWarnings about
missing keys indicate that `twine.__uri__` is being set to `None`.
`author` is also missing from package metadata.

This change iterates over Project-URLs looking for "Homepage",
and parses the author and email from Author-Email.
The email stdlib module is used for correctness; it is already
imported by importlib_metadata, so this does not add to import time.

changelog/1115.bugfix.rst

@@ -0,0 +1 @@
+Resolve DeprecationWarnings when extracting ``twine`` metadata.

twine/__init__.py

@@ -30,15 +30,20 @@
 
 __copyright__ = "Copyright 2019 Donald Stufft and individual contributors"
 
+import email
+
 import importlib_metadata
 
 metadata = importlib_metadata.metadata("twine")
 
 
 __title__ = metadata["name"]
 __summary__ = metadata["summary"]
-__uri__ = metadata["home-page"]
+__uri__ = next(
+    entry.split(", ")[1]
+    for entry in metadata.get_all("Project-URL", ())
+    if entry.startswith("Homepage")
+)
 __version__ = metadata["version"]
-__author__ = metadata["author"]
-__email__ = metadata["author-email"]
+__author__, __email__ = email.utils.parseaddr(metadata["author-email"])
 __license__ = None