It means that once you've been working for a given period of time you get a short period of (paid) time-off. A few lines of code in the scheduler to not assign work for 15 minutes to people in this position, and a corresponding credit in the billing system, would do.
The point applies more generally. I had a zero-hours contract with my university to teach very irregular hours on some IT training courses. I was entitled to holiday pay like every other worker. It accrued at a rate of something like one hour for every ten hours I worked, and was added to my pay.
This form of working isn't in any way new; piecework, home working, day labour and similar things have existed for as long as employment law has, and there are low friction solutions already in place to make meal breaks, holidays and minimum wage calculations work. The tribunal judges here specifically ruled on when Uber drivers are working (in licensed area with app switched on) so everything else is pretty straightforward from an implementation point of view.
That flexibility is nothing more than shifting risk from the employer, who is most able to handle it, to the employee, who is least able to handle it.
The positive I see in the so-called gig economy is far more flexibility in employing underutilised labour. If government increases the transaction costs of employing underutilised labour, then it will probably stop employing underutilised labour. It could easily be a net negative.
If there's an efficient way of working it out - i.e. keeping the transaction costs low - I'm not against regulating Uber as an employer. But it is a different form of working, and it's the kind of working that connected communication technology is best positioned to enable. It would be best if we didn't crush it under an assumption of regular employment.