OF-115 Remove race during outgoing session creation

OF-473: Revert original patch

Fix NPE in Node.java when changing node configuration.

Add and document LDAP startTls option to setup

Denote master version as 4.0.0 alpha

Also be more correct when reading the "xml:lang" attribute and remove superfluous whitespaces.

Fixes OF-977

https://community.igniterealtime.org/thread/48391
https://igniterealtime.org/issues/browse/OF-66Signed-off-by: Sebastian Wendel <swendel-git@sourceindex.de>

Referring to the patch from danm:
https://igniterealtime.org/issues/browse/OF-66

This NPE caused two issues:
- subscribers are not correctly notified
- change is not persisted to the DB

https://community.igniterealtime.org/thread/57584

Remove support for Legacy date/time (XEP-0090 and XEP-0091).

Delete web.xml.startup.servlet

Update admin.tld to use the latest taglib version.

Delete unused classes

This one is hard to duplicate in the lab, so this may not actually cure the
issue, however I suspect it does.

The existing code is based around a route of last resort, the
OutgoingSessionPromise, which builds a helper - the PacketsProcessor - and
creates the new session. The PacketsProcessors operate from within a thread
pool, and thus have contended execution time.

As new packets arrives from onward transmission during the session setup, the
PacketsProcessor queues them for retransmission. However, the act of session
creation also creates the outgoing route, meaning that later packets will
bypass the queue, and be transmitted out of order.

This should give the impression of a randomly reordered set of stanzas, but
would in fact be two consecutive sequences of ordered stanzas which are
interleaved, hence if the two sequences are A-E, F-J, one might see:

A, B, F, G, C, H, I, D, J, E

The fix here is two ensure that all packets are queued until the queue itself
is empty. In the solution provided, an additional helper class acts as the
route's RoutableChannelHandler, and stores packets in a queue until the real
ServerSession can catch up with the pending traffic. After that it simply
passes the process() call argument through.

OF-976 Language is not properly set in HttpSession

Language info should be in the Session, not in the Connection

Remove methods, which are both deprecated and unused.

Fix NPE when handling routing failures

Simplify and shorten pubsub message id

Update REST API plugin to 1.2.1

Avoids NPE by dropping (and optionally logging) error response stanzas
if the original packet has no sender JID.

The specification (XEP-0060) recommends generating unique message ids
within a pubsub instance to aid troubleshooting for notification
delivery issues when needed. This replaces a ridiculously long and
inefficient string concatenation with something a bit more tidy and
concise.

Fixed an cluster issue by create a chat room

Received reports of unexpected roster caching caused by this change (not
verified). Additional investigation is needed to determine whether/how
to fix original issue reported via OF-473.

These specifications are deprecated long enough (since 2007) in favor of XEP-202/203.

Extended the github readme

OF-115 Process S2S stanzas from a session serially

I feel the language (the stream's "xml:lang" attribute) should be in the Session interface rather than the Connection (similar as the stream id), because it's unrelated to the session's transport.

The HttpSession already did that.

Otherwise the used language is not easy to retrieve, e.g. when using SessionManager to get the session and OF-208 is hard to solve.

Alternatively we can expose the Connection in the Session interface, but this approach feels wrong to me.

Also use `java.util.Locale` rather than String, because it better reflects what can be in `xml:lang` (it can also have country codes, e.g. "de-CH")

Update web.xml schema namespaces to version 3.1 (Java EE 7)

Simplify/improve NIOConnection.deliver() by removing ByteBufferWriter.

For every packet being delivered the `ByteBufferWriter` creates a new String for every few chars written and puts it to the buffer.

According to my tests it is more efficient to create only one string and put it into the buffer once.
(10-20% better performance)