That makes zero sense. When the MCBs are closed, the capacitors are approximately shorted out. When an MCB opens, a small amount of current leaks into the circuit from another circuit via the capacitor(s). The left hand one appears to be marked 10N which I read as 10nF. Current into a dead circuit would be 230.2.pi.50.10exp-9=0.7mA. The only application I can see is to suppress a breaker opening transient due to some odd stuff being on the circuit, but then you would want the suppressors all to a common point, not from one circuit to the next. Another possibility is that something radiates RF and the circuits are acting as antennas, being mutually isolated at RF by the inductive reactance of the magnetic tripping coils. The capacitors shunt the tripping coils paralleling the circuits at RF. All a bit far fetched. I think it's generally a mistake.
They seem to be polycarbonate dielectric types (no longer available) of about 630V rating (from their size). Not being self-healing like class X and Y caps, they are not intended for mains suppression applications and could fail short-circuit and prevent an MCB disconnecting the circuit.
Hmm, there might be something to what Darkwood says about a joke. Somebody might have been trying to teach their apprentice a lesson about isolation. Put the caps in to leak a small current into a circuit, enough to give them a jolt or a funny reading but not likely to cause serious shock. The value is about right for that. But then didn't take them out afterwards. Or left it as an easter-egg for someone else.