First review pass

[caruizelembioinfo]

Addressed automated review comments:

Bugs

• cells2stats_bar_plots.py: Wrong logic for empty check - Fixed
• tabulate_manifest_stats: R2 description says R1 - Fixed
• Typo in tabulate_run_stats description - Fixed
• Trailing period in plot name - Fixed

Design Issues

• Random plot IDs break report reproducibility - Fixed (Removed random strings)
• find_log_files called twice to count files - Fixed (Single call collects log files and length is retrieved from list)
• Legacy data structures initialized but unused - Fixed (Removed ununsed legacy structures)
• Heavy code duplication between run-level and project-level parsers - Fixed (Created helper functions).

Code Quality

• Return type annotation mismatch in _parse_run_project_data - Fixed
• Potential KeyError in _setup_colors - Fixed, returns default.
• numpy import unused - Fixed
• natsort dependency - Checke presence in toml
• num_lines: 100 for RunManifest.json - Ignored (based on Phil's comment).
• Sorted order not preserved - Fixed (Using natsort consistently).

Minor / Style

• self.add_software_version(None) - No changes made since the call is performed after confirming run data was present.
• The in_project_sample_groups dict built in _setup_colors is always empty for non-project-level paths and gets merged in anyway — slightly misleading. Fixed (the dictionary was removed since even in project-level paths it had the same info as ind_sample_groups.
• search_patterns.yaml: the bases2fastq/manifest pattern uses contents: Settings and num_lines: 100. The Settings key in RunManifest.json appears near the top so 100 lines is probably fine, but worth a comment. Ignored.
• The module docstring in plot_runs.py and plot_samples.py remains as a bare string at module scope (not a real docstring) — per CLAUDE.md, module-level docstrings shouldn't be used; these can just be removed. Fixed (removed)

Missing Tests

• No tests are added for the new project-level parsing paths, the manifest parser, the index assignment parser, or the unassigned sequences parser. Given the complexity of the new code (especially the path-traversal logic in _parse_run_manifest_in_project), at minimum a fixture-based test with sample JSON files would help prevent regressions. Fixed (Added fixture-based tests and integration tests (pending merge in test-data)