I don't see many of your posts generic microbe but you did offer timely advice (if unseen) when I made that error a week or so ago, so I now get to see more of your posts. I also think the most likely explanation in this instance may be error but as has been already pointed out, the error rate is very, very small.
The BOM info sheet indicates that measurements are now 24 hourly, presumably automated.
Historic records would have been made with a max min thermometer, a utube shaped device with a sealed bulb of air on one side. A slug of mercury was free to move along the lower parts of the "U". This mercury pushed tiny bits of spring steel up the arms of the utube. The bits of steel were banana shaped so they would slide through the tube when pushed by the mercury but remain in place when the mercury retracted. In this way they provided a record of maximum and minimum positions for the mercury slug. Thermal expansion and contraction of the air in the bulb did the work.
The banana springs were reset with a magnet whenever readings were taken. Resetting both the minimum and maximum at 9am would mean that it would be physically impossible for the daylight maximum to be actually less than the previous nightly minimum imo.
With manually recorded temperature readings, occasional recording errors would be inevitable when literally millions of data points are being taken by hundreds of different people in many different locations over many years. The fact that the error rate is so small is testament to the dedication of the BOM staff. It is sad to see the work of these people being abused by socially corrupt, antiscience zealots.
