Implementation of Plotly viewers and refactoring

This PR takes a first step at addressing #3 by providing basic POC implementations of Plotly-based 2D scatter and histogram Jupyter viewers. Additionally, this PR contains some refactoring aimed at making as much of the code be frontend-agnostic as possible - in addition to powering Plotly viewers, it would also be nice to reuse much of what we have to make exporters for other frontends (i.e. bqplot). I started with the 2D scatter and histogram viewers because they're pretty standard visualizations, and we're interested in potentially using these with CosmicDS.

As of now, these viewers are a work in progress and undoubtedly still have some issues. The current state of things looks like this:

General viewers:

[x] Implement home/pan/zoom tools
[x] Implement ROI selection tools
[x] Decouple aspect ratio from axis range sizes

Scatter viewer:

[x] Basic visual attributes (color, size, colormaps, attribute scaling, opacity, fill, etc.)
[x] Lines
[ ] Vectors
[ ] Error bars
[x] Polar projection
[ ] Full-sphere projections

Histogram viewer:

[x] Basic visual attributes
[x] Normalized and cumulative bars
~~Normalized and cumulative settings work fine when set directly, but the limits don't come out right when they're set from the widget(?)~~ This seems to be an issue with the widget, as the same happens with the bqplot viewer

Generally speaking, most of the logic for representing these items in Plotly traces has been worked out in the static exporters, so it's just a matter of making these traces respond to updates in the state.

If someone is interested in trying these out:

Install this branch: pip install git+https://github.com/Carifio24/glue-plotly@viewers

Here's a sample snippet to run in a Jupyter notebook:

from glue.core import Data
from glue_jupyter import jglue
from glue_plotly.viewers.histogram.viewer import PlotlyHistogramView
from glue_plotly.viewers.scatter.viewer import PlotlyScatterView

app = jglue()
session = app.session
dc = app.session.data_collection

from random import randint
data = Data(label='data', x=[randint(0, 100) for _ in range(100)], y=[randint(0, 100) for _ in range(100)])
dc.append(data)

viewer = app.new_data_viewer(PlotlyScatterView, data=data)

glue-viz / glue-plotly

Implementation of Plotly viewers and refactoring #40