Okay, I was a bit optimistic there so it took a couple of days before I got to the bottom of this.

And surprisingly; looking at my data it seems that @pichlo is not completely out of the charts when he claims that v3.2.0.x would be more of a power-hog than previous releases.

This took me completely by surprise since it does not feel like that at all to me... but data doesn't lie.

Now, some grains of salt. The device in question is an XperiaX and my use patterns might have changed a bit during the months the data was gathered. YMMV.

So here's the charts since few last releases;

v3.0.2.8


v3.0.3.9


v3.1.0.11


v3.2.0.12


Every morning the charge starts from 100% and diminishes during the day, the different lines show different days.
(I just noticed that there are two "bumps" in the v3.1.0.11 chart on two lines, seems I put the device in charger for some time during those events.)

Charts are plotted on same normalized axis, the x runs from 0 to 200 minutes (33 hours) as power status is saved every 10 minutes.

Here are the data files for your covenience;
chartdata_v3.0.2.8.csv
chartdata_v3.0.3.9.csv
chartdata_v3.1.0.11.csv
chartdata_v3.2.0.12.csv