remove unnecessary lock in SharedMemoryResource::do_deallocate

[ShoroukRamzy]

Summary

This PR removes an unnecessary std::lock_guard from the SharedMemoryResource::do_deallocate method.

Problem

The SharedMemoryResource implements a monotonic (Arena) allocation strategy, where deallocate is a no-op. The current implementation acquires a heavyweight InterprocessMutex before doing nothing, which introduces unnecessary performance overhead on every deallocation call.

Solution

The std::lock_guard is removed from the function. Since no shared state is read or modified, no lock is required.

Benefits

• Performance: Eliminates a costly and unnecessary system call, making deallocation a true zero-cost operation as expected from a monotonic allocator.
• Clarity: The code now more accurately reflects the no-op nature of the function.

[github-actions]

The created documentation from the pull request is available at: docu-html