make sure addfeature is only called on readd

[jgravois]

resolves #886

[patrickarlt]

@jgravois I feel like we should prevent the problem (why is the event firing twice) rather then apply a solution (force feature to only be added once). I can try to find some time to debug why this happens.

[jgravois]

agreed.

I took a stab at that initially but i couldn't quite wrap my brain around the internal logic and sequence of events when a fresh filter is applied in a real world scenario.

when I updated an existing test to start counting the number of events fired I expected it to trap the bug the user reported, but the issue isn't reproducible when raw features are fed in manually via createLayers() at all.

[patrickarlt]

@jgravois Looks like @goldenlimit might have figured this out... #886 (comment)

[jgravois]

@goldenlimit was right.

the best course of action is to stop firing addfeature when layers are created entirely. this allows us to support the use case described in #788 and avoid the bug reported in #886.

since i never updated the API reference documentation to mention that addfeature was temporarily triggered when loading features for the first time, arguably what i'm now proposing isn't even a breaking change.