Leave GPU region unpinned by default

[kevintli]

Summary

Stacked on #3. BaseConfig.platform["modal"]["region"] now defaults to None, not "us-west", so Modal can schedule trainers and inference pools in any region. H200 capacity in a single region was the bottleneck.

# deployment_apps.py
trainer:  region=platform["modal"]["region"]          # None -> unpinned
rollout:  **server_placement(region)                   # {} when None, else routing_region+compute_region
# scoped.py: dropped hard-coded routing_region="us-west" (2 sites)

• Setting --region / platform.modal.region still pins both routing_region and compute_region, the same as before.
• If no region is set, app.server uses Modal's default routing_region (us-east). That setting only controls where the HTTP proxy ingress runs. compute_region=None leaves GPU placement unrestricted.
• Tests: region assertions were updated to match the new default. tests/providers, test_deployment_cli.py and test_deployments.py pass. The 4 failures in tests/scoped/test_lifecycle.py also fail on the tinker-025 branch without this change.
• Deployed from this branch for the sec-search-rl reproduction (Modal env kevinli-spindle-secrepro).

Link to Devin session: https://modal.devinenterprise.com/sessions/f53cfabb210146de8f0338fe388d7973
Open in Devin Desktop: https://modal.devinenterprise.com/desktop/session/f53cfabb210146de8f0338fe388d7973?variant=devin
Requested by: @kevintli

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access that start with 'Devin'.

• Disable automatic comment, CI, and merge conflict monitoring

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.