Seed: $500 starts, selectable AA and factory scaling reports

[TechLuddite]

New Seed runs start with $500, reducing the first $1,200 fab to seven manual chips. Existing saved balances remain unchanged.

The user reports factory scaling slowdowns on both laptop and phone, alongside jagged edges after MSAA was disabled. Add persistent graphics controls for off/2×/4× MSAA, resolution caps and shadows, while retaining the previous default GPU budget. Add an optional bounded local performance report with fab counts, frame intervals and rendering counters so comparisons can be made on affected devices. This does not claim the scaling problem is fixed.

Validation: JavaScript syntax and whitespace checks passed locally. GitHub Actions runs the simulation, original game and Chromium/Firefox checks, including new-start economics, save/graphics persistence, and recorded one/two/six-fab AA/shadow comparisons. No local browser, server or Godot runtime was launched, per the user's shutdown instruction.

All checks passed on final head ec94004 (Actions run 34579617726), including 16 Godot browser tests. Reports and screenshots are retained in the browser-checks artifact.

Short CI software-rendering comparison (1280×800, muted, same camera): one to six fabs increased draw calls from roughly 245 to 454 and rendered primitives from 97k to 198k; scene nodes stayed at 437 throughout. At six fabs, disabling shadows reduced draw calls to roughly 292 and primitives to 89k. AA was expensive on this software renderer. This supports investigating rendering cost as the factory scales, but does not establish the cause of the user's system-wide slowdown, rule out memory growth, or predict hardware-accelerated laptop/phone frame rates. Recordings include a few dialog-open frames around start/stop and are short functional comparisons, not controlled benchmarks.