On kbuild exception, store and send to API recorded stacktrace for easier troubleshooting and error reporting

Right now stacktrace is lost as soon as pod will be cleaned by k8s.
As half-solution kci-k8swatch.py from deploy will record logs, but it is not ideal.
We have already trap for exception, so maybe we can just add some basic function to send it somewhere, or schedule to send over another pod.