My take on bidirectional rcds at this stage is that unless the manufacturer states its bidirectional or its marked as unidirectional it's an unknown quantity
Indeed. Until very recently, I had always assumed that 'traditional' RCDs (without markings) were all 'unidirectional', and that 'bidirectional' was a relatively new-fangled concept which had arisen as a result of electricity export.
However, having just found a number of bog-standrad RCDs (RCCBs) on my shelves which are 20+ (quite possibly 25+) years old, which appear to be about as 'bidirectional' as they could be, I'm definitely coming to wonder whether there ever have been (at least, 'this century') any that were actually 'unilateral'.
As JohnW2 stated ,the test button problem may occur if the rcd is not operational , it can remain in-circuit if not designed to do so
Yes. Even with the additional switched contacts for the test button circuitry (which seems to be what is done to make a device 'bidirectional') there will
still be a potential problem with 'lengthy presses of the test button) if (due to being faulty) the device fails to trip when it should - and it would require a fair bit of additional complexity to address that possibility. As I've suggested, it may be felt that since such a device is 'already faulty', and hopefully will be replaced since it did not respond to the test button, that this is not a issue to worry about?
Also the electronics apparently may remain in circuit on the load side after tripping if there's 2 supplies if not designed to do so
Yes, but as I've said, if the device does trip when it is meant to, that is not a problem - since having the electronics permanently powered (for years/decades) is precisely what the devices are used to experiencing in-service.
A possible problem arises if (due to being faulty) the device fails to trip after the electronics has detected a residual current which should result in a trip (since it is not impossible that the electronics uses more current in that situation) -- but, as I have said (unless, I suppose, the devices became much more complex) I find it hard to see how that possibility could be addressed (just as with the 'test button problem'), even in a device said to be 'bidirectional'. If the device
does trip when it should, then any residual current will vanish, so that the electronics should return to their 'normal' state (in which being continuously powered is not a problem).