Browser test: avoid loading elements over network
Page object let us search elements for which attributes match a given
regex. Unfortunatly it loads the list of elements first, then foreach
of them trigger a call to Selenium to retrieve the attribute to match.
Attempting to regex match a link href, would cause hundred of roundtrip
to the Selenium host.
Instead, use a CSS selector to get the element we are interested in
which skips the bad page object code.