[Feature Request] Improve comparison algorithm to handle frame offsets intelligently (GSoC Qualification)

Inizia in mod_test/nicediff/diff.py, in particolare in get_html_diff intorno alle righe 123–217 e nella relativa generazione degli eventi. Leggi come l’attuale confronto riga per riga utilizza test_correct_lines e test_res_lines, quindi definisci il comportamento per gli intervalli di inserimento, eliminazione, sostituzione e uguaglianza. Il lavoro è completato quando i frame eliminati o spostati sono isolati, mentre le righe successive corrispondenti restano contrassegnate come superate e l’HTML diff è corretto.

Context

According to the "Sample Platform NG" GSoC project brief, one of the major goals is to improve the comparison algorithm. Currently, the comparison system is quite rigid, evaluating results strictly line-by-line. If a single frame (subtitle frame) is missing or slightly offset in the output, it causes a chain reaction where all subsequent frames are treated as a mismatch, causing the entire test to fail unnecessarily.

Current Behavior

In mod_test/nicediff/diff.py (around line 144 to 157 in get_html_diff), the logic iterates through use length and compares test_correct_lines[line] == test_res_lines[line].
This naive 1-to-1 index matching means that if test_res_lines drops line index 5, then index 6 of the result is compared against index 5 of the expected output, index 7 vs index 6, leading to a 100% mismatch from that point onwards.

Expected Behavior

The comparison algorithm should be smart enough to detect insertions, deletions, and offsets. If a single frame is dropped, the diff should highlight only that specific dropped frame as an error, and successfully resync/align the remaining lines that match.

Proposed Solution

I propose rewriting the core logic in mod_test/nicediff/diff.py (specifically get_html_diff and the event generation) to leverage a Sequence Matching algorithm (such as Python's built-in difflib.SequenceMatcher or a similar longest-common-subsequence approach).
This will allow us to:

1. Identify the exact opcode (replace, delete, insert, equal) for block ranges.
2. Gracefully handle offsets so the rest of the valid output remains marked as "Passed".
3. Render a much more accurate and readable HTML diff for the developers debugging the test suite.

Goal

I am applying for GSoC for this project under the mentorship of @carlos.fernandez and @canihavesomecoffee. I would love to take this up as my qualification task and submit a PR fixing this core comparison logic. Let me know if I have the green light to proceed with this!