* REMOVE second call to normalizeEmail() in
   EmailBasedDistributor.getBridgesForEmail(), and add note about where the
   checks are currently called.

 * ADD some extra logging when we receive imcoming requests from email
   addresses in the EMAIL_WHITELIST.

 * TODO check that incoming emails from addresses in EMAIL_WHITELIST have
   valid signatures made by the key with the corresponding GnuPG key
   fingerprint (as in EMAIL_WHITELIST, which is a mapping of addresses to
   fingerprints).

 * ADD bridgedb.Dist.EmailBasedDistributor.whitelist attribute which
   should hold the contents of the EMAIL_WHITELIST setting in
   bridgedb.conf.

 * CHANGE createBridgeRings() function in bridgedb.Main module to use
   the new email distributor whitelist attribute, and pass it the
   settings from the parsed config file.

 * CHANGE getMailTo() and runChecks() in
   b.e.autoresponder.SMTPAutoresponder to check for whitelisted
   addresses first, before any of the other checks.

Since there are technically two SMTP servers, be clearer that this
function sets up the one which listens for incoming emails.

When raising a `twisted.mail.smtp.SMTPBadSender`, we were giving it the
origin's domain name as argument, which ended up with weird messages
like '550 Cannot receive from specified address <example.com@localhost>:
Sender not acceptable' being sent out. Apparently, SMTPBadSender want
the whole origin object, not just the domain.

 * CHANGE raising of SMTPBadSender in
   b.e.server.SMTPIncomingDelivery.validateFrom() to use the whole
   origin object, not just the domain, as its argument.

If anything in this list ends up in the list of permitted client email
addresses, we should know about it, and also fail the test.

This serves two purposes. It allows arbitrary whitelisting of certain email
addresses for testing purposes, and once #9332 is fully implemented, will
allow encrypted emails to be sent to these whitelisted addresses.

 * FIXES part of #9332.

The stupid Python stdlib logging module doesn't handle unicode at
all. If debugging is turned on, and SAFELOGGING is disabled, the
incoming emails are logged. If they have any unicode characters,
processing of the emails stops due to UnicodeDecodeErrors.

Apparently, Transifex only displays comments which are for the string
being translated, and only one string is viewed at a time. Sigh.

 * THANKS to Phoul on IRC for pointing out the issue with how Transifex
   handles TRANSLATOR comments.

These are somehow the old strings from before the UIs were revised. It
seems the automated Transifex updater scripts didn't create a new PO
file for Korean… that or I did something wrong. Either way, we can still
use *some* of these strings, so we should add them.

 * THANKS to ilbe123, cwt96, Dr.what, and pCsOrI.

 * THANKS to LinuxChata and ghostishev.

 * THANKS to volkangezer.

 * THANKS to Communia, Humberto Sartini, Anastasia01, and recognitium.

 * THANKS to hoek, yodaa, maxxx, and sebx.

 * THANKS to Ann Boen, erwindelaat, guryman, and BBLN.

 * THANKS to plazmism.

 * THANKS to Random_R.

 * THANKS to Lunar.

 * THANKS to dark_yoshi and strel.

 * THANKS to pappasadrian.

 * THANKS to trantor and unknwon_anonymous.

 * THANKS to autofunk78 and DavidNielsen

 * FIXES #12122. This string wasn't getting extracted to the PO template
   file, lib/bridgedb/i18n/templates/bridgedb.pot.

 * FIXES part of #12122.

We don't need to test for conditions which might raise a
``twisted.protocols.tcp.CannotListenError``, we just need to ensure that
we exit if we can't bind to the port for the email server.

Some of them were taking a *really* long time on some hosts,
particularly these two:

   - test_SMTPAutoresponder_reply_transport_invalid
   - test_SMTPAutoresponder_reply_transport_valid