Configure submit recovery attempts

[Volv-G]

Summary

• extend failed-submit recovery lookup backoff to the full 0.5s, 1s, 2s, 4s, 8s, 16s schedule
• add configurable submit_recovery_attempts plumbing with OSS default 2 attempts
• expose --submit-recovery-attempts on tangle sdk pipeline-runs submit and route submits through the recovery-aware run path
• bump tangle-cli package version to 0.1.1

Context

Delayed Tangle run registration can make a successful submission look like a failed submit. The retry path already freezes the body and annotates a submission id, but short recovery lookups can miss runs that become searchable several seconds later, causing duplicate submissions. This keeps the conservative OSS default at two recovery lookups while allowing downstream deployments to opt into the full wait window.

Testing

• uv run pytest tests/test_pipeline_runs_cli.py -q
• uv run ruff check packages/tangle-cli/src/tangle_cli/__init__.py packages/tangle-cli/src/tangle_cli/pipeline_run_manager.py packages/tangle-cli/src/tangle_cli/pipeline_runner.py packages/tangle-cli/src/tangle_cli/pipeline_runs_cli.py tests/test_pipeline_runs_cli.py