The word "mean" is the same as the word "average", so when you see a reference to a "scaled mean" for a course, it's just a reference to the average scaled mark in that course.
To use your example, the average scaled mark in MEx1 was 36.6 on a 1-unit basis or 73.2 (36.6 * 2) on a 2-unit basis.
Because some courses are only worth one unit, and other courses might only contribute one unit to your UAI, the statistics relating to scaled marks are reported on a 1-unit basis.
The explanation for this question is a little more involved.
First, it is your raw HSC marks that are scaled, not the aligned marks. The reports include aligned marks because students usually do not know their raw marks.
You can see the general process from this flowchart:
http://www.boredofstudies.org/other/flowchart.pdf
The scaling process itself involves the simultaneous solution of a large set of linear equations. In essence, the scaling algorithm transforms all the raw HSC marks for each course into points on a scale between 0 and 50, with the gaps between points representing the differences between students. This way, the particular raw marks awarded don't matter, and only the relative positions of students in their courses are considered.
The scaled mean (or scaled average) for a course is defined to be the average scaled mark obtained by all students in that course over all of their courses. At the start of the scaling process, no one knows what these scaled means are going to be. The aim of the scaling algorithm is to produce a set of scaled means for all HSC courses which satisfies that definition.
The algorithm considers the positions of all students in all courses simultaneously and outputs a set of scaled means. Because of certain constraints specified in the algorithm, this set of scaled means is guaranteed to be a unique solution and is the only set of scaled means which satisfies the definition set out above as well as the constraints.
The raw marks in each course are then transformed to scaled marks using the scaled mean calculated for that course (as well as certain other data - e.g. scaled standard deviation and maximum possible scaled mark).
I've tried to keep this overview simple, because the mathematics involved is quite complicated, but hopefully it provides you with a better understanding of the process.