HTML repr for empty dataframes is ugly

see #5550 (comment)

related #6062

[TomAugspurger]

Let's keep an eye on jupyter/notebook#1182 and maybe close in favor of that?

[TomAugspurger]

xref: jupyterlab/jupyterlab#910 and specifically jupyterlab/jupyterlab#910 (comment)

Summary: Jupyter is taking care of a bunch of aesthetic issues in Jupyterlab. There are a few data-related items we can fix:

Ideally NaN would get styled slightly differently. That's particularly important for string column so you can distinguish missing values from the equivalent text.

Can add an extra class to NaN (maybe color, bold, or italicize?). Right now the HTML repr of 'NaN' the string literal and np.nan are identical

The fertility column has varying decimal places

This is python's string repr truncating 7.00 to just 7. I think we have another issue about that, will dig it up.

For columns like pop [large integer] you might want to consider adding commas every 3 digits.

This breaks copy-pastability, but thats pretty fragile, and we have df.to_clipboard() anyway.