René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Drop support for directly imread()ing urls.","articleBody":"### Describe the issue\r\n\r\n**Summary**\r\n\r\nSupport for directly imread()ing url strings caused Matplotlib to add a dependency on a package which we should not have no business of having a dependency on (`certifi`, see https://github.com/matplotlib/matplotlib/pull/18636#issuecomment-703126749).  url strings can't be directly `open`'ed either, so if we instead documented the (easy) solution of calling `urllib.request.urlopen` on the URL (+/- certifi) it'd be much more educative for users (now they know how to urlopen anything, not just images) (note that one of the main reasons for originally adding support to directly opening url strings was that the name of `urllib.request.urlopen` changed between Py2 and Py3, https://github.com/matplotlib/matplotlib/pull/4256#issuecomment-84369830, but that's obviously not relevant anymore).  Finally if we really want that functionality it would make much more sense to have it in Pillow anyways (given that imread is just a thin wrapper around PIL.Image.open at that point).\r\n\r\n### Proposed fix\r\n\r\nDeprecate and then drop support for `imread(\"https://...\")`.\r\n\r\n","author":{"url":"https://github.com/anntzer","@type":"Person","name":"anntzer"},"datePublished":"2020-10-03T16:17:14.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":0},"url":"https://github.com/18648/matplotlib/issues/18648"}