Drop FFMA (#435)

This PR drops FFMA (Fast-Fixed Memory Allocator), the novel memory
allocator implemented for cachegrand, because mimalloc has improved
enough to drammatically reduce the impact of the fragmentation as well
as allocating and deallocating memory buffer continuously.

To simplify the codebase, as the gap is now minimal, I have decided to
remove FFMA and use only one memory allocator instead of two.

README.md

@@ -65,10 +65,8 @@ interface and the WebAssembly support are still in the works.
   for monitoring
 - Blazing fast [hashtable](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/hashtable.md) capable
   of digesting 2.1 billion records per second on a 1x AMD EPYC 7502
-- [Fast Fixed Memory Allocator (or FFMA)](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/fast-fixed-memory-allocator.md) 
-  capable of allocating and freeing memory in O(1)
 - [Scales vertically](#benchmarks), 2x cpus means ~2x requests
-- In-memory and on-disk [timeseries database](https://github.com/danielealbano/cachegrand/blob/main/docs/architecture/timeseries-db.md) (WIP)
+- In-memory and on-disk storage
 - Supports different collection data types (Streams, Lists, Hashsets, Sorted Sets, etc.) (WIP)
 
 #### Planned Key Features

benches/bench-art-spsc-op-get.cpp

@@ -12,7 +12,6 @@
 #include "exttypes.h"
 #include "data_structures/double_linked_list/double_linked_list.h"
 #include "data_structures/queue_mpmc/queue_mpmc.h"
-#include "memory_allocator/ffma.h"
 #include "xalloc.h"
 
 #include "benchmark-program-simple.hpp"