omni: CI smoke + docs + README preamble

- .gitea/workflows/smoke.yml: gitea CI on ailab gpu runner (manual git
  clone since actions/checkout@v4 mis-resolves subpath gitea); injects
  WANDB_API_KEY + CI_RUN_TAG=smoke-$run_number
- scripts/smoke.sh: in-place smoke (uv sync + 1 shard + tokenizer +
  d=6 50-step base_train); idempotent cache at /data/nanochat-smoke/
- doc/research_feasibility.md: voice-first multimodal feasibility study (mochi)
- doc/todo.md: phase-by-phase roadmap (W1 Whisper smoke → W4 MVP)
- README.md: omni preamble pointing at upstream nanochat README
- .gitignore: exclude .claude/ runtime files

.gitea/workflows/smoke.yml

@@ -0,0 +1,25 @@
+name: smoke
+on:
+  push:
+  workflow_dispatch:
+
+jobs:
+  nanochat-smoke:
+    runs-on: gpu
+    timeout-minutes: 30
+    steps:
+      - name: checkout
+        # actions/checkout@v4 mis-resolves server_url on subpath gitea
+        # (drops /gitea/ prefix), so clone manually.
+        run: |
+          git init .
+          git remote add origin https://famzheng.me/gitea/${{ github.repository }}.git
+          git fetch --depth=1 origin ${{ github.sha }}
+          git checkout FETCH_HEAD
+      - name: nvidia-smi
+        run: nvidia-smi --query-gpu=name,memory.free,memory.used --format=csv
+      - name: smoke
+        env:
+          WANDB_API_KEY: ${{ secrets.WANDB_API_KEY }}
+          CI_RUN_TAG: smoke-${{ github.run_number }}
+        run: bash scripts/smoke.sh