Wow that's bad news about losing a bunch of source code! Sincere sympathies there - I can't imagine that happening! (Can't happen here, everything is locally backed up to another server and that backup is then replicated daily to Amazon S3). Indeed, the actual source code/project isn't even kept on the build machine's hard drive, it's accessed remotely from a server via iSCSI (which is running a raid 10 volume made up of a bunch of Samsung 840 SSD's). It's virtually impossible to lose more than say a day's worth even in the worse case scenario.
The reason WeatherCat won't start-up until it sees all primary sensors present and correct is to avoid a (common) problem where the user does some maintenance. They'll typically quit WeatherCat, put the console into maintenance mode (assuming a Davis station) then go do the maintenance. When done they boot WeatherCat and carry on with their life.
However, sometimes either a sensor has been left disconnected in the ISS or the console has been left in maintenance mode. The next day they realise no data is being recorded and get it fixed, the data downloads from the logger and everything is updated. Without this check they'd have been recording/uploading nothing and will very rapidly have an email into support asking how they can sort the mess out