Cache: WIP (4/5) Adopt inference caching in task runners

[ErlisLushtaku]

Description

Note

WIP: replacement cache stack on top of #118. Not final and may still change.

Adopts the inference cache in the task runners through --run.store_root.

• Pairwise, meta-eval, MT-Bench, and Elo generation and judging share the do_inference cache.
• Elo temperature calibration uses the same judgement cache.
• Cache rows get instruction, model pair, and orientation metadata at the direct inference call.
• Dataset-provided completions stay as direct inputs when they do not run inference.
• Removes ignore_cache, cache_function_dataframe, and the old cache tokens.

This is stacked on #123.