Investigate and add tests for potential memory leaks

We noticed some memory leaks in the codebase. While some fixes have been applied, we suspect not all issues have been addressed. To properly handle this, we need a structured investigation and tests that can reliably expose the leaks.

- https://github.com/blockblaz/zeam/pull/198
- https://github.com/blockblaz/zeam/pull/194

Next steps:
- Use memory profiling tools (Valgrind etc) to locate hotspots.
- Add a reproducible test case that highlights the leak(s).
- Investigate severity and prioritize fixes.

Start by reviewing pull requests 198 and 194 to understand which memory-leak fixes have already been applied. Use a memory-profiling tool such as Valgrind to locate remaining hotspots, then add a reproducible test that exposes a leak. Done means the leak is reproducible, its severity is assessed, and fixes are prioritized.

Written by the indexing model from the issue text.