Finish the implementation of Mark-and-Sweep Garbage Collection for the G-machine

[engesgaard]

Note: Changes related to this issue is located on the "gc_mark2" branch.

Sub-issues

• Avoiding the accumulation of orphans. This can be a result of application nodes being converted into indirection nodes of size 1. This can be remedied by making indirection nodes of size 2. However, this also requires us to make integer nodes of size 2, thus wasting more space on the heap.
• Avoiding that the garbage collector ends up in an infinite loop where there will never be freed enough space to finish the computation. This might be connected with the fact that the heap gets segmented. Then we need functionality to reorganize the heap to free larger blocks.

[engesgaard]

Both sub-issues have been solved, and the "gc_mark2" branch has been merged into "master". The first sub-issue related to orphans has been solved as described, by only allocating blocks of at least size 2. The second sub-issue related to segmentation of the heap has been solved by fusing together adjacent free blocks on the heap when sweeping.