tabicl_baseline: thread offload_mode/batch_size/disk_offload_dir through

[amrit110]

Summary

• TabICL comparator runs have only ever scored the basic 17-feature panel, never the strong 609-feature panel the GBM baseline uses, because a naive fit at the standard row/estimator settings costs an estimated ~70GB per predict_proba call and reliably OOM-killed the host (three confirmed kills, 2026-08-23).
• TabICLClassifier already exposes offload_mode/batch_size/disk_offload_dir to manage exactly this memory bottleneck, but none of it was wired through this project's tabicl_baseline.py. This threads those three params through fit_tabicl_baselines -> _fit_one_tabicl -> the classifier, and narrows check_inference_cost's RAM-budget gate to skip only for offload_mode="disk" with a real disk_offload_dir, since that combination changes the actual constraint from resident memory to disk I/O (which the existing GB estimate doesn't model, and shouldn't be trusted to block).
• Adds docs/tabicl_strong_feature_comparison.md: the first TabICL run on the strong feature panel this unblocks, with bootstrap CIs against the tuned GBM on all 12 MIMIC alert cells, plus an honest account of two failed configurations before a one-model-at-a-time loop completed cleanly.

Test plan

• uv run ruff check / uv run ruff format --check clean on both changed files
• uv run mypy odyssey/inference/tabicl_baseline.py clean
• uv run pytest tests/odyssey/inference/test_tabicl_baseline.py — 20/20 passed locally
• Empirically validated on the GPU host (odyssey-cbm-a100): all 12 (event, horizon) cells fit and scored without OOM using offload_mode="cpu", n_estimators=1, reduced row cap; see docs/tabicl_strong_feature_comparison.md for full results and caveats

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.03%. Comparing base (fd11be2) to head (7408591).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #213   +/-   ##
=======================================
  Coverage   89.03%   89.03%           
=======================================
  Files          57       57           
  Lines        8491     8493    +2     
=======================================
+ Hits         7560     7562    +2     
  Misses        931      931           

Files with missing lines	Coverage Δ
odyssey/inference/tabicl_baseline.py	98.03% <100.00%> (+0.03%)	⬆️

🚀 New features to boost your workflow:

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