ENH: Corrects to_html print spacing GH4987

[clham]

Closes #4987

This is my first non-doc PR, so please bare with me. Is it worth my while to go through the tests and standarize all of the test HTML to all block-strings, or is the mix of individual lines/blocks ok?

[jreback]

does this actually change the way the output looks?

[clham]

As far as I know it only strips off the leading/trailing spaces in the HTML data rows. Those spaces shouldn't be rendered by a browser?

[jreback]

@jorisvandenbossche @cpcloud @TomAugspurger look ok?

[jreback]

@clham can you add a release note? (put it under API changes section in whatsnew, just a single line and reference the underlying issue)

[clham]

@jreback, Should be good to go. I added a plain api section. Current version only
lists breakages.

[jreback]

@TomAugspurger @jorisvandenbossche any problem with this?

[jorisvandenbossche]

Seems OK to me

[jorisvandenbossche]

But I think this should not be listed in API changes, but can be considered as a bug fix.
And it needs a rebase.

[jreback]

merged via ed45e6e

thanks!

[jankatins]

Thanks a lot for this, I had some problems letting pandoc convert such html and was doing a re.sub(' +',' ', html)but after this it isn't nessesary anymore :-)