fix(ci): disable pip cache in build-wasm to prevent cache failures

Disabled setup-python pip caching that was causing persistent 400 errors
on every workflow run.

Root cause: Large ML packages (torch ~700MB-2GB, transformers ~500MB-1GB,
optimum, onnxruntime) create a 2-4GB pip cache that consistently fails
GitHub's cache service with 400 errors and service unavailability.

Impact:
- Removes cache service errors from workflow annotations
- Adds ~2-3 minutes to build time for pip install
- AI models output is still cached (no change to overall build time)

The AI models themselves are cached separately via actions/cache, so the
overall build time impact is minimal. Pip dependencies are fast to download
compared to the 10-15 minute model conversion process.

Author: jdalton

.github/workflows/build-wasm.yml

@@ -153,8 +153,11 @@ jobs:
         uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # v5.3.0
         with:
           python-version: '3.11'
-          cache: 'pip'
-          cache-dependency-path: 'packages/socketbin-cli-ai/requirements.txt'
+          # Disabled pip cache - large ML packages (torch, transformers) cause
+          # cache service failures (400 errors) on every run. Build time impact
+          # is minimal (~2-3 minutes) since AI models are cached separately.
+          # cache: 'pip'
+          # cache-dependency-path: 'packages/socketbin-cli-ai/requirements.txt'
 
       - name: Install Python dependencies
         run: |