Reduce shared memory implementation complexity

[jp-pino]

The shmem class is templated to take DATA_STRUCTURE classes that implement the concept of having a write and read functions and keep a storage type and state. The only implementation of this we have is the seq_lock. Unless we decide to ever implement an alternative shared memory container, complexity could be reduced to make the code more readable by making the shmem class only take in seq_lock. This way we wouldn't need the shmem_data_structure concept and there would be less templating on that class overall.