OF-1518: Ensure the summary is properly escaped

OF-1509: Add thread safety when incrementing counter.

* Release 2.3.0 of the Hazelcast plugin

* Correct the date

This updates the inVerse plugin to use the latest upstream release. As suggested by JC, it's author,
I've also disabled the 'viewMode' configuration option for now (this should be reenabled in future
releases).

OF-1345 Can't create/reset passwords when using HybridAuthProvider

OF-1421: Drop Java 7 support, require Java 8.

OF-1504: Caching the session will prevent carbons.

OF-1508: Prevent empty entity caps hash lookups.

* Updating the REST API Plugin to 1.3.0

Added: Security Audit endpoint to get the security logs
Added: More details by error
Improvement: Better JSON mapping (e.g. single element in an array bug)

* Adding missing libs for the ANT job

* Fixing the date for the plugin

2018 instead if 2017

OF-1507: Handle S2S stream errors.

OF-1191: Add BOSH mutual authentication config to admin console.

OF-1505: Prevent admin console stacktrace on TLS config page.

HZ-9: Ensure clustering works properly when enabled through the admin UI

OF-1495: Add subject alternative names to certs and CSRs

Depending on what components (plugins) are loaded, the certificate that is used to identify the local
domain should have additional values in its Subject Alternative Name extension.

This commit adds an information header in the TLS admin console pages when Openfire is serving components
that are not included in its certificate.

OF-1496 reduce discovery complexity

OF-1191: Use client truststores for c2s-type connections.

OF-1501: Prefer NewSunX509 KeyManager implementation.

OF-1481: Don't attempt to send over a non-existing connection.

Instead of waiting for data to become available on the main thread, this task is better offloaded to a dedicated
thread pool. This prevents worker threads from locking up (which potentially cripples the server).

This is a very basic (and inefficient, and prone to cause thread starvation) change that causes
MAM queries to be responded to only after all relevant data has become available in the database.

Prior to this commit, archived data was kept in a buffer, which was flushed to the database once per minute.
It is undesirable to have the database 'lag behind' for such an extend period of time.

This commit introduces near-instantanious flushes. Data that's put in a buffer is flushed immediately, unless
more data is available instantly.