Sync tree update

What's Changed

• Build apteryx-sync tree as a proper tree by @blairsteven in #38
• Small changes to sync_tree_process by @blairsteven in #39

Full Changelog: v2.07...v2.08

v2.07

What's Changed

• Fix generation of paths of synced variables by @tony-vanderpeet in #37

Full Changelog: v2.06...v2.07

v2.06: Fix the paths for the periodic sync update

The current code is inserting two slashes (//) into the periodic
sync update messages. This is causing issues on the receiving
device.

v2.05

Add support for syncing more complex paths

It is currently difficult to sync on a small portion of a model if that
model includes a list element. For example /node1/node2/*/node3/node4.
The current method would be to select node1/node2/* then exclude the
unwanted fields. This however can cause substantial processing.

This change allows paths of type /node1/node2/*/node3/node4 to be
entered. The new code uses apteryx_query and apteryx_watch_tree
to retrieve the database information to synchronize.

Additionally exclude paths are now only applied to the model they were
added with.

v2.04

Add mutex for delayed work list

The delayed_work list can be accessed by multiple threads which
can result in memory corruption.

Prevent this by protecting the access with a mutex.

v2.03

Take instance lock before destroying delayed work object

When the delayed work object is freed the call reference is removed from
the registry index - this work requires the state to be locked.

v2.02

Fix memory issue when dpulicate entry in map file

The insertion of a map entry was mishandled when there is a
duplicate entry. The library routine actually takes care of
the situation by itself.

v2.01

Only load apteryx API for XML files

v2.00

Multithreaded Alfred

With new lua instance locking functionality in Apteryx we can now load
lua files and lua snippets from XML files into seperate lua stacks.

This requires changes to the lua that is loaded - currently each lua
file in the script directory will be loaded into its own lua state, and
the XML files will all be loaded into a single state.

v1.99

apteryx-saver: Check for apteryx_query() failure whilst writing config

Check for apteryx_query() failure whilst writing config. Apteryx_query()
can return NULL if a refresher path included in the query doesn't
respond to apteryxd.