Feature: optimize map rendering

[Fur4nk]

Summary

This PR introduces significant browser performance optimizations for the Crucix dashboard, focusing on resource management and user control. Key changes include:

• On-demand 3D Engine: Deferring the loading of three.js and globe.gl until the 3D mode is manually activated.
• WebGL Lifecycle Management: Implementing pauseAnimation() when switching to Flat mode to eliminate idle GPU/CPU usage.
• Persistent Low Performance Mode: A new "PERF" toggle (saved in localStorage) that disables expensive CSS backdrop-filter effects, halts purely aesthetic animations, and simplifies globe features
  (reduced star field, thinner atmosphere).
• Improved Ticker Accessibility: In Low Perf mode, tickers switch from auto-scrolling to manual scrollable lists with custom themed scrollbars for Chromium/Firefox.
• Symbology Fixes: Removed priority-based culling; all markers now remain visible at all zoom levels on both projections.

Why

The previous implementation caused high CPU/GPU load, particularly in Firefox, due to the constant background rendering of the 3D globe and the intensive use of CSS blur filters. This led to excessive fan
noise and poor responsiveness on mid-range hardware. Users also reported "missing" data points at world-view zoom due to priority-based filtering.

Scope

• Focused bug fix
• Small UX improvement
• New source
• Dashboard change
• Docs/config change

Validation

• Manual testing of the "PERF" toggle: verified backdrop-filter removal and animation halt via Browser DevTools.
• Verified WebGL loop suspension in Flat mode using chrome://tracing and Firefox Performance tab.
• Confirmed localStorage persistence across page reloads.
• Visual verification of custom scrollbars on Chromium (Edge/Chrome) and Firefox.
• Zoomed out to world-view on both projections to confirm all markers (Air, Thermal, OSINT, etc.) remain visible.

Screenshots

• New "PERF" toggle in the top-right status bar.
• Low Performance Mode active: clean panels (no blur), static/scrollable news lists.
• 3D Globe with simplified atmosphere and optimized star field.

Config and Docs

• No new environment variables
• .env.example updated if needed
• README.md updated if behavior changed

Source Additions

N/A

Checklist

• This PR stays within one bugfix or one feature family
• I kept unrelated changes out of the diff

[calesthio]

I like the direction here, but I found three correctness issues in the current head:

1. toggleFlights() still assumes globe exists. Because this PR defers globe initialization, clicking the flight toggle while still in the default flat view can hit globe.arcsData(...) / globe.pointsData(...) on null.

2. Low-perf mode is not actually sticky for globe arc animation. togglePerfMode() disables dash animation, but onGlobeZoom() re-enables it whenever the camera gets close enough. That means a low-perf user can zoom back into one of the expensive animation paths the feature is supposed to suppress.

3. The globe load-failure path is unsafe. initGlobe() can catch a script-load/init failure and leave globe unset, but toggleMapMode() still unconditionally calls globe.resumeAnimation() afterward. If CDN loading fails, switching to globe mode will still throw.

The performance goal is valid, but I don’t think this is merge-ready until those null/error paths are closed.

[Fur4nk]

You were right. I hadn’t tested the related paths thoroughly enough...

I went back through the branch and fixed each of the problems:

• toggleFlights() no longer assumes the globe exists, so toggling flights while still in flat mode or before globe init is safe
• low-perf mode is now sticky for globe arc behavior, including zoom changes, and I also made sure auto-rotate does not resume in low-perf after user interaction
• the globe load-failure path is now guarded properly, so switching to globe mode won’t call resumeAnimation() if globe initialization failed

While retesting, I also cleaned up a related issue in the lazy globe bootstrap path to avoid duplicate Three.js loads.

I also rebased the branch onto current master. A few related changes had already moved upstream in the meantime, so I cleaned up the overlaps during the rebase and kept only the remaining branch-specific
pieces.

Sorry as well for the branch churn. I realized I had been committing with my work Git identity, so I had to rewrite the branch history to correct the author/email before pushing the updated branch. That part
is on me.

[calesthio]

I haven't been able to get to this yet because work has been busy, but I definitely plan to review it over the weekend.