repo sync OWNER/REPO, run where no Git remote points to the repository,
now updates it on the server: a fork's branch fast-forwards to its
parent's through merge-upstream, and a pull mirror fetches its source.
repo fork reports an existing fork ("! bob/x already exists") instead of
a raw HTTP 409, refuses to fork a repository into its own owner, and
gains --fork-name, --clone (Git flags after --) and --remote, which adds
the fork as origin and renames the old origin to upstream.
repo clone passes Git flags after --, and takes --upstream-remote-name
and --no-upstream for forks. Clones report what they set up.
repo set-default checks that the repository exists before saving it,
prints the saved repository with --view, removes it with --unset, and
reports the change.
Signed-off-by: bircni <bircni@icloud.com>
tea repo sync ran git switch on the synced branch, so syncing main while
on experiment left the user on main. A branch that is not checked out now
moves with update-ref, guarded by its old value; the checked-out branch
merges with --ff-only and needs a clean working tree; a missing branch is
created and set to track the remote. Diverged branches stay untouched.
Sync reports "✓ Synced main from origin/main (fast-forward a..b)" on a
terminal, "already up to date" when nothing changed, and the JSON result
gains the previous commit. Outside a clone, or without a remote for the
repository, the errors now name repo sync instead of checkout.
Signed-off-by: bircni <bircni@icloud.com>