Add document loaded/addedForStorage interception methods

[tim-cools]

I added two methods to the IDocumentSessionListener to intercept when documents are loaded/stored.

void DocumentLoaded(object id, object document);
void DocumentAddedForStorage(object id, object document);

This interception is currently implemented in the default and DirtyChecking IdentityMap.

I was tempted to split the IDocumentSessionListener into a IDocumentSessionListener and IDocumentListener. The latter would contain the new methods. But I decided to go for the simplest implementation for now. Let me know what your thoughts are about this?

If you agree I can add this also to the documentation if you want.

[ekulakov]

@tim-cools Maybe I'm wrong but I think you could write a decorator and reuse it for all IdentityMap's

    public class IdentityMapListenersDecorator : IIdentityMap
    {
        private readonly IIdentityMap _identityMap;
        private readonly IEnumerable<IListener> _listeners;

        public IdentityMapListenersDecorator(IIdentityMap identityMap, IEnumerable<IListener> listeners)
        {
            _identityMap = identityMap;
            _listeners = listeners;
        }

        public T Get<T>(object id, Func<FetchResult<T>> result) where T : class
        {
            var obj = _identityMap.Get(id, result);
            _listeners.Each(x => x.DocumentLoaded(id, obj));
            return obj;
        }

        ...
    }

[tim-cools]

@ekulakov thanks for the feedback!This is however not possible because the listeners are only called when a document is loaded the first time and not when the document is returned from the cache.

tim-cools@306bb17#diff-99cac1569732f4ed1d27aceb9e8f52ebR64

A decorator is unaware of the internals of the decorated class. We might consider pulling out a generic base class for both IdentityMaps however. I will look into that later...

[jeremydmiller]

@tim-cools I'd rather we used a "nullo" empty list here instead of having to remember to always use the "?" operator. Nit pick-y, I know.

[tim-cools]

no problem, will update it in a next pull request. it was a silly attempt of premature optimization...

[jeremydmiller]

I've got a minor comment about using a nullo for the list of listeners, but other than that I think this looks good. The build passed, and I'm pulling this in now.