2.0.0-rc isPending briefly returns false during overlapping updates to an async memo chain

[GabbeV]

Describe the bug

Updating a signal twice while its async memo chain is pending makes isPending report true → false → true → false.

Both writes happen before the first false, and no further writes happen when it becomes true again.

Your Example Website or App

https://s.olid.uk/id/_vddZGrFTleyH3BBtMLRbg

Steps to Reproduce the Bug or Issue

1. Wait for the initial load.
2. Click Run once.
3. Watch Pending briefly become false before becoming true again.

Expected behavior

Pending should stay true until the replacement update finishes.

Screenshots or Videos

No response

Platform

• OS: Windows
• Browser: Firefox

Additional context

The same case with longer delays was verified on next at 344ed05 and #3337 at 20f3143.

Replacing the two async memos with one async memo of the same total duration removes the false interval.

[ryansolid]

Reproduced and fixed in #3381 — same root cause as #3373.

b's first flight lands while a has been re-asked by the second write. The landing cleared every pending entry on b (including the one a's re-ask propagated), so for the gap between that landing and a's second landing b read as settled — isPending(b) blipped false → true. With the landing retiring only b's own entry, b stays pending on a and the verdict stays true until b's final answer: false → true → false.

— Claude via Cursor

[ryansolid]

Fixed on next in d7cb456 (#3381). Same root cause as #3373: the first-hop landing wiped the propagated pending entry for the re-asked input, so isPending(b) read false for the gap. The landing now retires only the node's own entry and b stays pending on a until a's second flight lands.

— Claude via Cursor