A bit of an explanation of 'how it should function' - now I've dug out the correct circuit diagram!
There are two RF receivers, one active (button press, connects to the KVM by a serial data connection from the antenna module {top edge of tailgate IIRC}) this is the lock/unlock activate interior sensor and trigger alarm as required. Communication is on the MS-CAN. The other passive system is the immobiliser which works on a push-pull NFC system, this is the one where you need to place the key against the dash to re-sync the codes and connects directly to the CJB, which in turn broadcasts the status on HS-CAN.
One of the differences between MS & HS-CAN is that MS-CAN can operate 'single-legged', that is, should you lose one side of the pair of wires to damage, corrosion or a short to ground or +ve, the other will continue to broadcast, so it's actually quite a resilient communication method. HS-CAN requires (mainly due to the greater bandwidth & error detection / rejection) both 'legs' to be operational, therefore the transceivers will shut off either side of the fault, but continue to operate either side of it, where possible, some modules will also keep working as 'standalone' controllers, generally with reduced performance or abilities.
Now down to the reason for the above explanation - it's very unlikely to be either network due to the amount of data that would be potentially corrupted with an unbalanced MS-CAN fault, there would be several DTC's set, although the integrity of that network is only monitored for a full open/short type event. So the remote unlock problem is likely to be receiver power, data, or ground, KVM problem (unlikely as no DTC set) or CJB (possible as there is an unexplained / unusual CCF issue) Likewise, the immobiliser issue also relates to the CJB in terms of physical connections but also processing of received signals and onward transmission.
Intermittent issues would indicate an internal defect with the hardware being affected by temperature or moisture, it's possible I suppose (not my area of experience) that a failing processor and memory in the CJB would also affect how it processes inputs & outputs if the coding is corrupted. As usual with 'black box' modules, you have to test the input & output devices and wiring for correct operation in order to condemn or otherwise the suspected module (CJB in this case). It's not an easy thing to do where you're looking for variations in CAN & Serial signals to indicate reaction to inputs - it doesn't matter what the signal 'says', you're just looking for a reaction to input.