chore(deps): update unilab-rl to 1.4.9

[TATP-233]

Summary

• Update the uni_rl optional extra and dev dependency group from unilab-rl==1.4.8 to 1.4.9.
• Refresh uv.lock with the published 1.4.9 registry artifacts.
• Make three off-policy terminal logger tests explicitly select the 120-column layout they assert. unilab-rl 1.4.9 now chooses its layout from the logger-owned Rich console width; without a TTY that console falls back to 80 columns, so the prior tests depended on the runner environment.

Linked Work

• Issue: Work: update unilab-rl to 1.4.9 #2087
• Parent roadmap (when applicable): N/A
• Roadmap declared base (when applicable): N/A
• Milestone: N/A
• Base branch: main

Validation

• make test-all passed on the final local head before this PR was created
• Additional task-specific validation listed below

Commands actually run:

uv lock
# Updated unilab-rl v1.4.8 -> v1.4.9

uv sync --extra mujoco --extra uni_rl
# installed unilab-rl==1.4.9

uv run --no-sync python -c 'import uni_rl; print(uni_rl.__version__)'
# 1.4.9

uv run --no-sync pytest tests/test_library_import_boundary.py tests/ipc/test_dp_launcher.py tests/base/backend/test_process_device.py tests/algos/test_appo_runner.py
# 82 passed, 3 deselected

uv run --no-sync pytest tests/utils/test_experiment_tracking.py
# initial run: 3 failed under the non-TTY 80-column console
# final run: 23 passed

make check
# passed

make test
# 1821 passed, 43 skipped, 489 deselected

make test-all
# passed
# 1821 passed, 43 skipped, 489 deselected
# benchmark smoke: 34/34 module imports, 35/35 script imports

Remote CI route:

• Base main: CI passed on head 867388d5.

Impact

• Backend impact: none
• Platform impact: both (dependency resolution and terminal tests)
• Training effect expected: no (dependency/runtime adoption; 1.4.9 terminal rendering adapts to console width)

Artifacts

• W&B: N/A
• benchmark result: benchmark smoke output included in validation
• video / screenshot: N/A
• ONNX / checkpoint: N/A

Checklist

• Added or updated tests where needed
• Updated docs if behavior or workflow changed
• Linked the driving issue
• Noted any follow-up work explicitly (none)