refactor: modernize mutex locking from C++11 to C++17 (std::scoped_lock)

[Erio-Harrison]

Summary

Modernize all mutex locking from C++11's std::lock_guard to C++17's std::scoped_lock.

Changes

Commit 1: Dual-mutex scenarios

Before (C++11 - manual deadlock avoidance):

std::lock(mutex_, other.mutex_);
std::lock_guard<std::mutex> lock1(mutex_, std::adopt_lock);
std::lock_guard<std::mutex> lock2(other.mutex_, std::adopt_lock);

After (C++17 - automatic deadlock avoidance):

std::scoped_lock lock(mutex_, other.mutex_);

Commit 2: Single-mutex scenarios + headers + examples

Before (C++11):

std::lock_guard<std::mutex> lock(mutex_);

After (C++17):

std::scoped_lock lock(mutex_);

references

cppreference: std::scoped_lock
Stack Overflow: std::lock_guard or std::scoped_lock?
SonarSource RSPEC-5997

[jsulmont]

Hey! Thanks for the contribution — the scoped_lock modernization looks good and I'll merge it.

Quick note for future contributions: the standard OSS workflow is:

1. Fork the repo
2. Create a feature branch on your fork (e.g., feature/scoped-lock)
3. Open a PR from your-fork/feature-branch → upstream/main
4. After merge, sync your fork's main with upstream

Committing directly to your fork's main makes it harder to keep your fork in sync and creates merge noise. Feature branches keep things clean.

No big deal this time — just something to keep in mind!