Through the SpikeDetector Editor, it's now possible to change the channel
mapping for each electrode and adjust thresholds. It's not yet possible
to inactivate particular electrodes or change parameters over multiple
electrodes at once; these will both be necessary functions in the future.

Now, when the user clicks on a tab in the DataViewport, the corresponding Editor becomes
active. Likewise, when the user selects an Editor with a Visualizer, the corresponding
tab becomes active in the DataViewport.

Channel selection now takes place within a slick new interface. It's not
yet tied to actually changing parameters, but things should be much simpler
code-wise, as well as visually.

Added method for creating parameter editors, which can be overridden by subclasses of the GenericEditor

I commented out the activateAntiAliasing call and added a call to disableAntiAliasing for the SpikeDisplayCanvas as antialiasing messes up the projection plots. We can revert this change in the future if we come up with a way to enable anti-aliasing without completely messing up the projection plots

Slightly altered the layout of the SpikeDisplayEditor, also I changed the waveform line width from 1 to 2 pixels

added check to ProjectionAxes::createFBO() such that glDeleteFramebuffers and glDeleteRenderbuffers are only called if the FBO and RBO have already been generated. OpenGL specs indicate it is fine to call these functions on non existant handles, like on the initial creation of the texture. This wasn't a problem on linux but it was for OSX. It was causing each projection plot to share a single texture and all spikes were going to a single plot. As when the 2nd projection plot called createFBO it deleted the FBO for projection plot 1 and 3 did it for 2, resulting in a single fbo for all plots.  This should resolve most of the projection plotting issues for mac

fixed mipmap generation bug for OSX. Previous versions of OpenGl wanted you to hint to generate mipmaps, OSX wants you to explicity call the function that generates them.

The AudioComponent now closes the audio device when it's not in use,
saving CPU cycles. Previously, background CPU levels would be between
8 and 10%. Now, when acquisition is not active (and audio device settings
are not being actively edited), background CPU levels drop to 2%. More
optimization will be necessary to get it down to zero, but 2% is still
a big improvement over what we had previously.