feat(local-dev): add --json status output and a non-TTY build heartbeat

[Yicong-Huang]

What changes were proposed in this PR?

Two small, agent/script-friendly additions to bin/local-dev.sh — no change to the human TTY experience.

1. --json machine-readable status. status --json prints one JSON object on stdout (no colours, no table) and exits 0 iff every service is running, else 1:

{"branch":"...","sha":"...","running":14,"total":14,"services":[
  {"service":"texera-web","port":8080,"type":"jvm","pid":25823,"state":"running"}, ...]}

up and down also accept --json: human progress is routed to stderr (unbuffered) and the final status JSON goes to stdout (via a saved fd), so a caller can up --json >state.json 2>progress.log and parse stdout directly.

2. Non-TTY build heartbeat. In non-TTY mode the spinner can't render in place, so a long silent step (sbt dist, output redirected to a log) used to print one line then go quiet for 25s+ — indistinguishable from "stuck" to a non-interactive caller. tui_spinner now emits … still running (Ns) every TUI_HEARTBEAT_SECS (default 15), polling at 1s so it still returns within ~1s of the job finishing (no trailing latency).

Any related issues, documentation, discussions?

Closes #6026. Usage banner (--help) updated to document --json on status/up/down.

How was this PR tested?

• bash bin/local-dev/tests/test_local_dev_sh.sh → 19 passed, 0 failed (added 6: JSON shape/consistency, health-based exit code, unknown-flag negative case, --help coverage, heartbeat regression guard, up/down --json wiring).
• python -m pytest bin/local-dev/tests/ → 38 passed (no regression).
• Dogfooded end to end: down --json → stdout pure JSON (running:0, 14 stopped), exit 0; up --json → single-line JSON running:14/14 on stdout, and stderr showed the new heartbeat across a ~60s build (… still running (15s/30s/45s/60s)), 14/14 healthy.
• Verified stdout/stderr separation: up --json >state.json 2>progress.log yields parseable JSON in state.json.

Was this PR authored or co-authored using generative AI tooling?

Generated-by: Claude Code (Opus 4.8)

[github-actions]

Automated Reviewer Suggestions

Based on the git blame history of the changed files, we recommend the following reviewers:

• No candidates found from git blame history.