An update on this one: This ended up being related to the FunCube frequency setting as defined on the SensorStation. In this case, the frequency was set to the actual frequency for this region (150.1) and not the actual frequency minus 4 kHz (150.096). The distinction may seem somewhat trivial but if affects the frequency offset as recorded in the raw data. This in turn caused the recorded frequency offset to exceed the threshold permitted during processing. The tagfinder that runs on the SG or SS observes no such threshold (nor does it account for burst interval) which is why the tags were showing up locally, but not in the processed data. We were able to reprocess the previous data for this receiver with the frequency offset parameter corrected so the earlier detections appeared.
Note this the issue discussed here is the same one.