v0.1.14

Added

• The node announces its in-flight jobs on connect, so a hub redeploy adopts work already under way
  rather than restarting your stage. (#58)

Fixed

• Restarting the node mid-stage resumes from the on-disk job list (~/.dahrk/jobs.json) instead of
  re-running the stage from scratch. (#58)
• An interrupted stage preserves the killed agent's work on a dahrk/wip/<runId> ref and resets the
  worktree to the last complete commit. (#58)