While testing the updated test, I find out that it spent most of its time in `chunk_coalesc' before actually reaching the out of memory problem. I observed this because after the 7941th iteration it ran much slower (about 1s per iteration whereas it did the first 7941 in less than 6s) until reaching iteration 8291 where it got the exception out of memory and printed the stack trace immediately and exited immdiately.

After I removed the call to `chunk_coalesc' done via `full_coalesc_unsafe' in `malloc_free_list' then the program executes quickly and fails quickly.

So I'm not sure if we are observing the same behavior here.