fix(save): guard atomic-saver commit callbacks against a killed buffer

The per-field atomic savers advance the baseline and rewrite the display through the marker (org-entry-put marker / org-with-point-at marker) inside the async push callback. If the buffer is killed before the push lands, those touch a dead marker and signal. The deleted pearl--push-issue-field carried this guard, but it went away with the immediate-push setters and the atomic engine never picked it up.

pearl--run-atomic-field-save now skips the commit when the marker's buffer is dead (a new pearl--marker-live-p, which also passes a nil marker for synthetic callers) and still reports the remote outcome -- the push already landed, so only the local buffer update is moot. All three commit sites are guarded: the clean push, the converged-remote case, and the use-remote conflict resolution.

0 files changed, 0 insertions, 0 deletions