New hook invoked after all nodes have been rebuilt

Reroll attached.

Edit: this patch has a bug - see below for newer version

My last upload had a bug. Attached is the proper patch.

Rerolling this patch has actually turned out to be more difficult that I thought. Implementations of the "rebuild_complete" hook can make changes to the nodes and call node_save (such as #1041932: Export node translations does). This causes a problem with nodes appearing as overridden due to timestamps - something that was fixed in #1008496: Node timestamps, options overridden and cause uuuid_node components to appear modified.

The attached patch moves #1008496's logic that updates node timestamps into it's own function. This makes uuid_node_features_rebuild() easier to read, and also allows it to handle the node timestamp updates *after* implementations of hook_uuid_node_features_rebuild_complete() are called.

Could someone please review and flag RTBC?

@ryan_courtnage haven't been able to fully test this, but the db_update() in _uuid_node_features_update_node_values() should wrap $table in double braces for database prefixes:

db_query("UPDATE {{$table}} SET $column = %d WHERE $where", $args);

also Ryan, I'm using your patch in #4 in combo with #1048470-4: Book export, and almost everything is working… Except have you noticed a regression of #1008496: Node timestamps, options overridden and cause uuuid_node components to appear modified with this patch? I'm now seeing 'changed', 'date', 'revision_timestamp', and 'revision_uuid' in my features diff – not sure where that's coming from yet

@scottrigby, regarding:

_uuid_node_features_update_node_values() should wrap $table in double braces for database prefixes

Not sure I've ever seen this done before. Why double braces?

@scottrigby, in response to #7, I am not having problems with timestamps in my features diff. Are you using my patch from #4 (uuid_features_rebuilt_hook-1049760-4.patch), unmodified?

The Book Export patch/comment you reference calls for the patch in comment #1 by chaps2. That patch (uuid_features-1049760.patch) will most certainly cause the timestamp problems you are seeing, as explained in my comment #4.

I've just introduced this or better a similar hook for #1048470-15: Book export
The current implementation doesn't deal with timestamps yet. I've first to get familiar with the issue,
However, this definitely needs a re-roll anyway, so back to needs work ;)