Desktop Dev Build #191.1 (532024b)

Automated desktop build artifacts for commit 532024bfdbda6fa4667e9202f4d78717ea766ad6.
Workflow run: https://github.com/LettuceAI/app/actions/runs/28627210697

Changes since previous Desktop build desktop-dev-189-1-439d713

Compare: desktop-dev-189-1-439d713...532024b

• 532024b fix(sprout): validate hardware schema version
• 7538b12 fix(navigation): skip settings root on desktop back
• 3351f3d fix(navigation): correct mobile settings back flow
• 07ee146 docs(changelog): add mobile browser entries to 2.1.0
• 0414b3d feat(i18n): translate mobile ollama strings for all locales
• b459ebb feat(installed-models): reorganize page for mobile with ollama focus
• dd01587 feat(hf-browser): mobile ollama-only browsing with files drawer and compact controls
• 8ebc395 feat(hf-browser): redesign destination picker and persist ollama selection
• 97e53cb docs(changelog): reword multi-gpu entry as smart offloader
• 42aae64 docs(changelog): add 2.1.0 in-app changelog entry
• bc2f3a3 fix(embedding): install missing v4 tokenizer once on 2.0.0 upgrade
• 1e1a379 feat(i18n): translate memory cycle hub strings for all locales
• 2061a8e fix(memory): preserve embedding migration markers through session schema
• eb57a84 feat(export): sync chat export/import with official SillyTavern jsonl format
• 215e625 feat(memory): add shared memory cycle hub to chat and group memories pages
• 31ce4b1 fix(memory): retry windows after dynamic-memory failures
• 9eeadf1 feat(openrouter): add per-model provider pinning
• 114a9e4 fix: preserve companion mode data in UEC app settings
• d3915f7 fix: preserve current data in backup and sync
• 660cfd9 fix: preserve explicit dynamic memory model selection
• 26deb85 fix(llama): fix multi-GPU VRAM planning, layer estimation, and key al… (#70)
• 1f1362f i18n: translate Sprout runnability and provider strings
• c60ae28 feat: use Sprout hardware probe for remote Ollama runnability