How do I get the HTML in an element using Capybara?

You can call HTML DOM innerHTML Property:

find('//tr[2]//td[7]')['innerHTML']

Should work for any browser or driver.
You can check all available properties on w3schools