Allocation mechanisms improvements

[m4drat]

• ❓ Change allocation logic (firstly reuse all available space in free list, and only then allocate new arena). This one is under a question mark, because with that implementation we are losing benefits of simple bump allocator. But the idea in general is that we should use free-list more.
• ✔ Try to reuse the most appropriate chunks from chunk treap first (GetFirstGreaterOrEqualToChunk -> upper_bound). To do so it is a great idea to implement iterator for ChunkTreap #39 (just to use std::lower_bound)
• ✔ Before doing anything, complete Performance comparisons (mimalloc, malloc, jemalloc) + cpu/memory Profiling #29 and collect some base measurements.