2 seconds x 10 passengers (say) x 20 stops = 400 seconds, or six and a half minutes[1]. It all adds up - and might mean an extra vehicle on the PVR. It always amazes me how bus operators in the UK don't seem to be bothered by this (and similar time losses by having single-door vehicles, and having to chat to the driver about your destination rather than just touch in and out). The ultimate aim has to be that boarding passengers don't even break stride on boarding - that's how it works on TfL, and (traffic aside) the progress of London buses can be much quicker as a result - it gets very near to matching the speed of conductor operation.
[1] I've often pointed out the comparable situation that adding a hustle alarm to the doors on Merseyrail EMUs necessitated adding two minutes to the Ormskirk-Liverpool running time. With lots of stops, a couple of seconds here and there really add up.
I've switched to m-ticket recently to save money and find the biggest delay (after people paying cash) is waiting for ticketer to register peoples m-tickets. Takes a lot longer than it does to scan a printed 3d barcode for some reason.
It surprises me that Ticketer have used traditional laser barcode scanners rather than a camera. The former often have trouble scanning phone screens, require the brightness to be whacked up etc - and some simply won't work at all, for instance I can't scan a Nectar card or Clubcard from my phone (a recent Motorola with a decent LCD screen) on a Tesco/Sainsburys self service till. Whereas a camera with LED light can almost always read from literally anything. (FWIW, the problems using laser barcode scanners to read from phone screens is one reason why Parkrun require you to print your barcode - the other one, as an aside, being so that you have printed "in case of emergency" info on your person).
Another slightly odd "feature" is the way they have the contactless/pass reader under the barcode reader, which can apparently cause issues reading from phones with NFC switched on - the Arriva app tells you to switch it off, which is a bit of a cheap and nasty workaround to designing things properly. The TfL "Oyster pad" target seems to be a better design to me.