Optimize the Query component matching.

[AndreaCatania]

Improves the query component matching, so it's possible to exclude entities in batch.

Consider the case where there are two group of entities, that both have in common 1 component:

• Entity group A: Transform, Polygon
• Entity group B: Transform, Shape

Consider we have the following queries:

Query<Changed<Transform>, Polygon>
Query<Changed<Transform>, Shape>

Now, let's suppose we have 1 million entities, where half are Polygons; and let's suppose we change the polygons transforms.

As is now, the query that fetches the polygons will run fine, but the query with Shapes will have to fetch one by one its own transforms to see if they changed, and this is an useless check since nothing is changed: however, this check may become trivial, if we could exclude the polygons transforms from that query, we would know pretty easily that there is nothing to check.

A way to optimize this use case, would be to implement something like the Hibitset or a grouping mechanism.