Ensure bench results in vtcomposite are comparable or better than `node-mapnik'

[artemp]

#34
spent some time digging into bench results as they are confusing. I think I understand now why node-mapnik is x100 faster when overzooming two tiles with the same layer ('polygons') than vtcomposite. (edited)

node-mapnik doesn't do any work in this case and just returns input tile, vtcomposite on the other hand builds output tile from input tile. We need to introduce similar to node-mapnik short-cut logic in vtcomposite - on my TODO. Meanwhile, it would be great to test overzooming on input with different layers e.g polygons-1 and polygons-2 to trigger some real work in node-mapnik 🙂
/cc @millzpaugh @mapsam

[artemp]

To get comparable results we need to use add_existing_layer or similar approach. Currently failing sanitized builds #37

[artemp]

#42

[millzpaugh]

Comparisons are now accurate as of: #77