Enhance project performance consider cherry picking

[firestrand]

This PR is mostly to share the ideas and let others see the performance benefits.
The changes are around optimizing the QR rendering and frame order to minimize the size of the file.
The results are:

• Content accuracy preserved 100%
• Storage efficiency improves by 65.7%
• Performance increases by 61%

On top of the frame reordering updated:

• QR_BOX_SIZE from 5 to 3
• QR_BORDER from 3 to 1

There are several scripts in examples for seeing what is going on:

1. examples/concurrent_performance_demo.py - new concurrent frame odering measures multi-threaded vs single-threaded speed ups.
2. examples/early_stopping_demo.py - demonstrates early stopping logic that cots the progressive resolution loop short when further gains are negligible
3. examples/scaling_benchmarks.py - scaling across datasets from a few dozen to thousands of frames
4. examples/compression_benchmark.py - end to end and apples to apples compression tests summarizing the size savings from frame ordering overhead
5. examples/test_phase3_integration_demo.py - full integration of frame ordering into the build_video() including retrieval accuracy checks and error handling examples
6. examples/test_frame_ordering_demo.py - walkthrough of the progressive resolution sorting algorithm through encoder integration up to the latest features and analyses