feat: update invlerp and transferFunction heuristic for emitIntensity

[seankmartin]

Currently if you have more than one invlerp or transfer function UI control and use max projection volume rendering, if no emitIntensity is manually called in the user defined shader with neuroglancer, neuroglancer will use the value of the last called invlerp or transfer function to set the intensity used during max projection rendering.

While we can't always guess exactly what should be the emitIntensity with multiple invlerps it might be a more reasonable heuristic to take the max across all the invlerps / transferFunctions instead of the current behaviour. This would also need some tweaking for what to do when the mode is min instead of max.

Opening this to discuss that possibility. Will need to update some documentation about the rendering if this heuristic is used instead of the current one. cc @fcollman as this came up from an example

I guess the alternative is that maybe this wouldn't come up as a problem if we can adjust the blending across multiple channels when each channel is it's own layer

[seankmartin]

Closing to reduce noise. Ideal fix is to improve blending and avoid these kind of shaders needing to be written.

If it comes to it, within the shader itself the user can take a min/max over channels and emitIntensity instead of relying on the default.