WIP: (1/7) Add content-addressed local inference caches

[ErlisLushtaku]

Description

Note

WIP: This cache stack is on hold until the task-YAML stack ending in #93 merges. The current implementation is not final and may change when it is adapted to the merged task runtime.

Adds content-addressed local SQLite caches for model completions and judge outputs.

Cache layout:

{store_root}/completions/{task}/{provider}/{model}/{descriptor_hash}/
├── metadata.json
└── completions.db

{store_root}/judgements/{task}/{provider}/{model}/{descriptor_hash}/
├── metadata.json
└── judgements.db

metadata.json stores the validated model descriptor. The databases use these schemas:

completions(
  input_hash TEXT PRIMARY KEY,
  input_text TEXT NOT NULL,
  completion TEXT NOT NULL,
  benchmark TEXT NOT NULL,
  instruction_id TEXT NOT NULL,
  model TEXT NOT NULL,
  pushed_by TEXT NOT NULL,
  pushed_at TEXT NOT NULL,
  run_id TEXT NOT NULL
)

judgements(
  input_hash TEXT PRIMARY KEY,
  judge_input TEXT NOT NULL,
  judge_completion TEXT NOT NULL,
  benchmark TEXT NOT NULL,
  instruction_id TEXT NOT NULL,
  model_a TEXT NOT NULL,
  model_b TEXT NOT NULL,
  judge TEXT NOT NULL,
  orientation TEXT,
  pushed_by TEXT NOT NULL,
  pushed_at TEXT NOT NULL,
  run_id TEXT NOT NULL
)

This is stacked on #71.