Split all the e2e tests in their own files

Some duplication is avoided, but most importantly: it’s much much cleaner,
easier to find, functions have been cleaned up (functools.partial usages are
now hidden), etc.

tests/end_to_end/scenarios/irc_server_connection_failure.py

@@ -0,0 +1,12 @@
+from scenarios import *
+
+scenario = (
+    sequences.handshake(),
+    send_stanza("<presence from='{jid_one}/{resource_one}' to='#foo%doesnotexist@{biboumi_host}/{nick_one}' />"),
+    expect_stanza("/message/body[text()='Connecting to doesnotexist:6697 (encrypted)']"),
+    expect_stanza("/message/body[re:test(text(), 'Connection failed: (Domain name not found|Name or service not known)')]"),
+    expect_stanza("/presence[@from='#foo%doesnotexist@{biboumi_host}/{nick_one}']/muc:x",
+                  "/presence/error[@type='cancel']/stanza:item-not-found",
+                  "/presence/error[@type='cancel']/stanza:text[re:test(text(), '(Domain name not found|Name or service not known)')]",
+                  ),
+)