Add support for Plotly

[Valdes-Tresanco-MS]

Plotly is a fairly robust web plotting tool. Advantage:

• Generate very cool graphics, both static images, and interactive graphics.
• Allows saving in the same formats as matplotlib
• Includes similar color palettes as seaborn
• Complex graphics can be made (e.g. heatmap with categories in rows and columns)
• Works with pandas Dataframes
• Has graphics not implemented natively in matplotlib/seaborn
• Render well
• Easy to customize
• Easy to update charts with new data
• [Update] With hardware acceleration it is ~7.5 times faster than matplotlib

Disadvantages:

• Being interactive, as well as the need to launch multiple QtWebEngine threads makes it consume quite a lot of RAM resources. [Update] 50-90MB per plot, depending on chart type and data size
• For small data, it works very well, but when the data exceeds 1000 it requires hardware acceleration to run smoothly and render well. [Update] It seems not to be a problem. See the description in more details below

Other technical questions:

• I need to get familiar with Plotly
• Implement it as an alternative to seaborn and matplotlib, until making the relevant performance and functionality comparisons. [Update] Given the characteristics and differences with matplotlib, it will definitely be an option, although it is probably the default option. See the description in more details below
• Implement all current functionalities

I think it can be implemented for version 2.0

Recommended by Dr. Carlos Alessandro Fuzo

[Valdes-Tresanco-MS]

Some points related to the disadvantages: Each graphic is rendered in an HTML interface, so several QtWebEngineProcess (4) are created that consume about 50-90 MBs of RAM. This means that it will consume 50-90 MBs more than matplotlib representing the same graph. Rendering doesn't seem to be a problem, although I have to do tests on less powerful hardware. If you have a GPU that supports WebGL2.0, then it works extremely well, if not, use the CPU (I need to test it) as matplotlib

Based on my tests with a df of 200,000 frames, it draws the figure in just 0.73 s (using hardware acceleration), while seaborn/matplotlib takes about 5.5s. Its implementation is very worthwhile, the user will be the one who decides what he wants: a fast, efficient, interactive graphic, but with greater RAM resources or a slower drawing, little interaction, and less RAM