That is an interesting one and something I am not sure about in terms of what BS5839 says. Its clear that the C&E system will need to be tested and if you can do it silenty it might be easier Why not ask the good folk at this forum for advice - but do come back and share what you have decided http://firealarmengineers.com/forum/index.php

Good luck

To be honest, I cannot recall what the relevant BS says about testing a C&E program. But one I was involved with was so complex, it had to be monitored an tested regularly 

I designed a C&E (with others) that dealt with a phased evacuation with a staff search/time delay across 18 floors split into 36 evacuation zones.

There were 6 different voice alarm messages that were broadcast to the 36 zones as decided by which head had activated (heat, sprinkler or smoke) and where the head was located . 

The C&E had to deal with progressing the evacuation when an additional head was operated or if another floor became involved. The operation of a manual call point also progressed the phased evac, but only in some areas.

Linked to the alarm system the C&E had to deal with 24 lifts (6 of them fire-fighter lifts), 3 commercial kitchen extracts, 6 pressurised staircases, forced atrium venting, fire door closers, mag lock releases, automatic illuminated directional signs, beacons and goodness knows what else including a complex BMS system

We also has a water mist and several inergen systems feeding the C&E

It wasamongst  the most difficult days of my fire safety career trying to determine the various permiations and I  nearly had  a breakdown working on the biggest paper spreadsheet I have ever seen. I literally went home with a splitting headache several times, especially after I got it wrong and it failed on the first three trials.

We had a sophisticated monitoring system that kept an eye on the C&E software 24/7 but even then we did six month silent C&E tests 

So perhaps you need to risk assess how complex the C&E is and develop a testing regime to suit