feat(tabpfn): served model id follows the loaded checkpoint (0.11.0)

[amrit110]

Summary

• TabPFNBackend.name was hardcoded to tabpfn-3, but tabpfn 9.x loads TabPFN-3.5 by default (a different model and license). A default install therefore served TabPFN-3.5 named tabpfn-3.
• The id is now derived from tabpfn's resolved model_version (TABPFN_MODEL_VERSION): v3.5 → tabpfn-3.5, v3 → tabpfn-3, v3.5-fast → tabpfn-3.5-fast.
• The pretraining-limit integration test now exceeds the 160-class limit. The feature limit it used is 20,000 on v3.5 (2,000 on v3), so it no longer tripped.
• Bumps to 0.11.0. Needed by VectorInstitute/inference-platform#168 to serve TabPFN-3.5 in prod.

Breaking: callers that send "model": "tabpfn-3" to a deployment on the default version must switch to tabpfn-3.5, or the deployment must pin TABPFN_MODEL_VERSION=v3.

Test plan

• Unit suite (259) + ruff
• tests/integration/test_tabpfn_backend.py on CPU with tabpfn==9.1.0, offline (HF_HUB_OFFLINE=1, no token, read-only cache dir): 4/4 on v3 and 4/4 on v3.5
• build_engine() with TABCTX_BACKEND=tabicl,tabpfn serves ['tabicl-v2', 'tabpfn-3.5'] (v3.5) and ['tabicl-v2', 'tabpfn-3'] (v3)

🤖 Generated with Claude Code

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.50%. Comparing base (021f6e5) to head (af6fa7b).

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #29   +/-   ##
=======================================
  Coverage   89.50%   89.50%           
=======================================
  Files          25       25           
  Lines        1372     1372           
=======================================
  Hits         1228     1228           
  Misses        144      144           

Flag	Coverage Δ
integration	39.79% <ø> (ø)
unit	82.06% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/tabctx/serve/app.py	42.14% <ø> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.