do not clear _currentSnapshot when hiding the layer

[gavinr]

The behavior in #1207 is happening because the first time the layer is hidden due to the maxZoom, this._currentSnapshot is emptied, so the next time it tries to hide the layers, this._currentSnapshot is empty so it does not remove anything.

The bug is fixed if you remove that line (this PR), but I'm not quite convinced this is the correct fix because I'm guessing there's some side-effect that requires that line, and I'm just not understanding it - @patrickarlt @jgravois please let me know if you have any insights into why this line (this._currentSnapshot = [];) was originally included.

[jgravois]

🤷 nobody has touched that code in 5 years.

off the top of my head it doesn't seem problematic to preserve the cache instead of destroying it in the block.

the only other thing i'd suggest testing would be applying a new filter both while the layer is at a visible zoom level and when its not. i have a very vague recollection of folks complaining about a stale cache in a scenario like that.

[gavinr]

Thanks @jgravois - you were correct there was an issue when zooming in, calling setWhere, then zooming out/in.

Seems like in _requestFeatures function second parameter should be an object (correct here), but in three other places the code was passing a string to _requestFeatures(), which was causing issues. Fixed that in 87b4d14. (affects setWhere(), setTimeRange(), and refresh())

(note this might address the issues from #665)

[patrickarlt]

@gavinr this looks like a good change. Lets merge this.

[gavinr]

thanks Patrick. @jwasilgeo found one small issue - i'm going to try to fix that quickly before merge.

[jwasilgeo]

@gavinr, I'd like to also add the Labeling Features demo page to our list of manual tests. The odd behavior described in #1193 was fixed as a byproduct of v2.4.0 and I'd like to close it out if it still works as expected with the changes in this PR.

[jwasilgeo]

Everything looks great. Thanks for diving into this one and fixing it, @gavinr.