[codex] Update pyBer analysis and UI tools

[Andrianarivelo]

Summary

This update publishes the recent pyBer analysis and UI work:

• restores and extends Signal Event Analyzer peak detection, metrics, overlays, CSV/H5 persistence, and MAD-based auto transient detection
• adds baseline-prominence normalized amplitude with MAD fallback when baseline peaks are unavailable
• adds a noise trace / MAD threshold overlay for peak detection diagnostics
• adds temporal modeling UI and GLM/FLMM workspace panels
• improves numeric setting controls with arrowless draggable fields and adaptive decimal stepping
• makes app dark/light theme more portable by forcing Fusion palette styling independent of host OS theme
• expands processed output/export handling used by the updated workflows

Why

The previous analyzer could fail to show peak overlays/metrics, normalized prominence could return nan when the baseline window had no qualifying peaks, and some machines showed inconsistent dark-mode controls because native Qt/Windows styling leaked through the stylesheet.

Validation

• python -m compileall -q pyBer
• synthetic signal analyzer smoke tests for auto-MAD detection and normalized amplitude fallback
• temporal modeling widget/GLM smoke test
• full app smoke launch with PYBER_SMOKE_TEST=1 and QT_QPA_PLATFORM=offscreen
• PyInstaller rebuild from pyBer_build_clean
• packaged dist\pyBer.exe smoke launch

Notes

Generated __pycache__ files and local .vscode machine-specific settings were intentionally not included.