Explore texture-based approach for 2D datasets

[TaylorMutch]

Rendering dense datasets can run into video memory limitations. It may be useful to instead perform a texture-based approach to rendering out large datasets.

[TaylorMutch]

A texture-based approach affords us two possible advantages, and also potentially poses a challenge.

Advantages:

• We can handle the coloring of the textures with shaders. Meshes can provide single float-wide data buffer (i.e. Terrain & Color) that is then handled with either a continuous, binned, or categorical fragment shader (the vertex shader can be uniform across all 3). These fragment shaders could also be associated with a given instance of a Legend (or a shader inherits from Legend, or takes a Legend as an argument - ShaderProgram.create_from_legend(legend_instance), etc.)
• We can bring in openly available basemaps (OpenStreetMaps, Carto, etc.)

Disadvantages:

• Would need to add a hook for exporting time-series datasets to alert the Exporter for when a dataset has finished rendering all of it's tile meshes.

Considerations:

• Take good care for how blending between multiple layers occurs.

Overall, #34 and/or #22 could benefit from delving into this. It would allow for users to have more flexibility in terms of how their visualization looks.

[TaylorMutch]

May be useful in solving rendering issues with #41.