Maybe someone can chime in with a more definitive answer, but my understanding is that this is due to an upstream limitation in the way collectd and rrdtool calculate the min and max values. They are not actually an absolute min/max over different time scales.
That said, I don’t think this should be the case in SCALE as it no longer uses collectd for reporting.
My understanding here is that a wider timeframe used a lower number of samples to represent the data so as not to cause UI performance issues when loading the graph, so I never considered it a bug. This is based purely on assumption though.