PR #814 follow-ups: --model in spawn_dev's docs and its closing report nf-306d ← Beads

Twinned from https://github.com/cfsh/ninefold/issues/816 by tools/beads/import_github.py's reconcile pass.

Suggestions from ninefold-reviewer's review of #814. Both are advisory residuals — nothing in #814 is blocked on them, and either is cheap enough to fold into that PR before it merges.

1. --model appears in no synopsis. tools/office/spawn_dev.py's module docstring usage block (:4-10) and tools/office/README.md's "Spawning a dev session" block (:27-33) between them name every other flag the tool has — --issue/--notes, --adopt, --kickoff-file, --resume, --standing-slot, --dry-run. --model is the first one represented in neither, so a CTO reading the docs rather than --help will not find it. notes/office.md names tools/office/README.md as the rule "written for a reader standing in that directory", and that README says the module docstring "is the rest" — those two files are the tool's documented surface, and no test gates either. - Related: tools/office/README.md's step 4 spells the launch line as claude --remote-control <name> --permission-mode auto "<kickoff>", which is now one of two shapes.

1. The closing report does not echo the model. report() in spawn_dev.py prints the worktree, the tmux session, the adopted PR, the box and the lanes — the facts the CTO checks back against what they asked for. The model is now one of those and is invisible outside --dry-run: a spawn with --model opus when sonnet was meant launches cleanly and reports nothing that would say so. A typo is already caught — claude refuses a bad alias, dies at once, and launch()'s liveness check refuses noisily — so this is only the valid-but-wrong case.