Efficient Memory Management with PyMem_Raw Allocators in Python's _ssl and _hashlib Modules

Memory management in programming languages involves allocating and freeing memory efficiently to optimize performance and resource usage. In Python's C extensions, using appropriate allocators like PyMem_Raw* helps manage memory more directly and with less overhead, especially when interfacing with external libraries like OpenSSL. This approach reduces memory bloat and fragmentation by avoiding unnecessary Python object tracking for raw buffers.

This PR refactors the _ssl and _hashlib modules to use PyMem_Raw* allocators for OpenSSL-related memory allocations instead of higher-level Python allocators. By switching to PyMem_RawMalloc and PyMem_RawFree, the code bypasses Python's memory management overhead for raw OpenSSL contexts and buffers. The diff shows a reduction in memory usage during SSL context creation and EVP context initialization, as confirmed by tracemalloc measurements, demonstrating that this change effectively reduces memory consumption without altering functionality.

Understanding and applying efficient memory management techniques like using raw allocators can significantly reduce the memory footprint of critical cryptographic operations, leading to better application performance and scalability, especially in security-sensitive and resource-constrained environments.

Review the changes made to the _ssl module in this PR. How would you identify other parts of the Python standard library or third-party C extensions where switching to PyMem_Raw* allocators could yield similar memory savings? Propose a strategy for safely auditing and refactoring such code.