The original browser event was used to catch the mousemove event,
but in IE the 'pointermove' event is returned instead. So, instead
of using the original event, we use the map browser event, which
is always 'pointermove'.
Making it possible to choose between three select interactions, one that works on "singleclick", one that works on "click", and one that works on "mousemove".
Previously, if a layer didn't have a `getFeature*ForPixel` type method, the loop count would be decremented. This means that all layers would not be considered. Instead, the callback count should be incremented before requests are made and decremented in the callback.
A 'this' identifier is quite useless with compressed JavaScript,
and in fact it turned out to fail in advanced mode when trying
to access a feature's renderIntent property with it. The added
renderIntent lib function as a Call expression does the job
well.
With this change, the user provides a filter function instead of
an array of layers. Selection layers are created lazily, and
addition/removal of layers is not handled by the control to give
the user more options, as suggested by @elemoine.
