I guess technically you just found a bug, but it's not a big one.
Like blueskull already said, this behaviour of the 1054Z is normal if you have the trigger mode set to "Normal" or "Single", but it doesn't make sense in "Auto" (if you set a large time base).
So what's going on? If you set the time base to 1 s/div the 1054Z doesn't go into roll mode yet, even though it kinda looks like the roll mode (but the roll mode doesn't have trigger). And if you have your trigger point in the middle of the screen, then at 1 s/div this means you have 6 s to the left (pre-trigger) and 6 s to the right (post-trigger). In "Normal" mode the scope will wait and fill up memory for those 6 s to the left before it does anything, because that's pre trigger information it needs so it can display a complete 6 s of waveform to the left of the trigger point when a trigger occurs. After it got 6 s it continuous to capture pre-trigger information and is now armed to accept a trigger. Once a trigger occurs the last 6 s of waveform get displayed and the remaining 6 s to the right are drawn as they come. So that's perfectly fine behaviour, because Rigol doesn't want to display incomplete waveforms, so you gotta wait until you got all the pre-trigger first.
Now in "Auto" it doesn't make sense (for such a large time base), because during the 6 s of pre-trigger capture the scope can't be triggered, and once it got 6 s worth of information it will very quickly do an auto display of the waveform anyway, so why not draw the pre-trigger capture instantaneously as well. I would say this is only an issue for the small window of time bases, where the scope doesn't switch to roll mode yet and on the other side where the update rate isn't bigger than the display refresh rate yet.
So if this annoys you I would suggest to either switch to roll mode in the horizontal menu, or move the trigger point to the left of the screens, so there is no pre-trigger capture.