* fix: prevent false empty states during stack hydration
Only show confirmed-empty UI after successful stack, status, and container
fetches. Distinguish loading and recoverable error states in the sidebar,
dashboard, and container health panel.
* fix: arbitrate overlapping stack status and container fetches
Prevent older dashboard status and same-owner container responses from overwriting newer load state after concurrent poll, invalidation, retry, or lifecycle refresh.
* fix: do not let soft status polls starve slow foreground loads
Skip soft /stacks/statuses poll and invalidation while a statuses request is already in flight so a deferred foreground hydration can still commit after the ten-second cadence.
* fix(stacks): surface recoverable errors for confirmed-empty soft failures
Sidebar and dashboard soft (background) refresh failures after a
confirmed-empty state silently kept showing the empty/adopt prompt
instead of a recoverable error, since only the error message was set
without flipping the load status. Also reject malformed non-array
/stacks responses instead of coercing them into a confirmed-empty list,
and drop malformed per-stack status entries before they reach the
dashboard table, which previously crashed the entire app on a null
entry.
* fix(stacks): close two review-found gaps in the load-failure fix
A non-empty stack-statuses map where every entry failed validation was
still committed as a confirmed-empty success; it now surfaces as a
recoverable error instead, and dropped entries are logged. The sidebar's
background-failure helper also checked a stale closure snapshot of the
file list, which could wipe a list that had just loaded non-empty in the
same attempt if the follow-up statuses fetch then failed; it now tracks
the freshest committed list for that decision. Also collapses two refs
tracking dashboard status-map emptiness into one.
Anso
b06dfd7175