resetlisted

For a run that went wrong early. The triage read the issue as something it is not, the scouts mapped the wrong package, a plan was approved that should not have been — and continuing means every later step inherits the mistake, because every later step reads what the earlier ones wrote. This forgets one issue so the next full cycle starts from nothing. ## Steps 1. **`pdkit reset <issue>`** — the dry run. Nothing is written. It prints what would go, and, below it, what would stay. 2. **Read the "stays" half out loud to the user.** It is the half people skip and the only one that can surprise them: - **an open pull request stays open.** This forgets it; it does not close it. If starting over means the published attempt should not stand, that is a separate, human action on GitHub — and the plugin cannot do it anyway. - **the journal keeps every entry.** It is append-only by construction, so the earlier attempt stays legible to whoever asks in six months. Without that, an issue back at `new` would be indistinguishable from one nobody ever touched. - **deferrals survive**, for the same reason: a promise made to a reviewer is not undone by us deciding to start again locally. 3. **`pdkit reset <issue> --confirm`.** The artefacts are archived rather than deleted, under `$PDKIT_HOME/archive/<issue>/<timestamp>/`; nothing reads them, and `mv` puts them back. `--purge` deletes instead, for the case where the point is that the p