lakebox: resolve names → IDs from local cache

[akshaysingla-db]

databricks lakebox ssh my-project and the same shape on status / stop / start / delete / config / default now accept a --name value in place of a sandbox ID, resolved purely from local state — no extra API call on the slow path.

How it works:

• The state file (~/.databricks/lakebox.json) grows a per-profile sandboxes list of (id, name) pairs.
• resolveLocalID maps the user-typed arg to an ID:
  1. Exact ID match → return as-is (fast path; preserves the existing "I know the ID" workflow).
  2. Exact name match in cache → return the corresponding ID. Ambiguous names (server allows duplicate --name values) error out with the candidate IDs.
  3. No match → pass arg through unchanged. Either it's an ID the cache hasn't seen yet (fresh sandbox, user hasn't listed) or a typo; either way the next API call surfaces a clean 404.

The cache is populated incrementally by the commands that already observe sandbox state:

• create upserts the new sandbox.
• list replaces the cache in full.
• status / stop / start / config / default upsert the one entry they touch.
• delete removes it.

A stale cache can only fail to find a name; it can never cause the CLI to operate on the wrong sandbox. Resolution is ID-first so a sandbox whose --name happens to collide with another sandbox's ID never gets mistakenly matched by name.

Co-authored-by: Isaac

Changes

Why

Tests

[github-actions]

Waiting for approval

Based on git history, these people are best suited to review:

• @pietern -- recent work in cmd/lakebox/

Eligible reviewers: @andrewnester, @anton-107, @denik, @renaudhartert-db, @shreyas-goenka, @simonfaltum

_{Suggestions based on git history. See OWNERS for ownership rules.}