Moat's first IP comes from moat-shim, as this removes any other
X-Forwarded-From headers. And the last one is comming from the
connection between the moat-shim and apache, so this one is the public
IP address of the moat-shim server.

* Closes: #231

Thanks to knoagnel for finding the bug.

Upartitioned resources should be sent to all recipients that have
requested that type.

Let's make more clear of what the recipients are receiving.

Updating the same slice that is being used in the for produces
situations where some elements might not be removed. Let's do the remove
of the elements outside the loop.

Before this change there was one event recipient request with all the
recipient options per distributor. So if two streams were started with
different requests only the first one will be used. Let's have a request
configuration per channel, so we can have multipl streams with different
options on the same distributor.

Let distributors decide the IP version they want to subscribe to.

* Closes: #221

So we return a proper error in the HTTPS distributor and don't report
into metrics gibberish transports.

* Closes: #227

Let's make it easier to debug emails not being responded.

* Closes: #224

meskio authored Sep 02, 2024 and onyinyang committed Sep 04, 2024

defers are executed in oposite direction that then are called. If we
call the close defer after the unregistration of the channel it will
executed first. It can happen that an event arrive between the two of
them and crash rdsys.

* Closes: #222

HTTPS and moat distributors know the country from the IP address, let's
not distribute bridges known to be blocked in that country.

* Closes: #204

* Closes: #216