fix(nanoviews): show the for_ placeholder when there is no array at all

[dangreen]

for_ over a signal that holds null or undefined threw:

TypeError: Cannot read properties of null (reading 'length')

Both on the first render and on a change from a list to nothing. The static form already did the right thing — it happens to be written $items?.length ? … : else_?.() — so the two forms disagreed about the same input.

No array is the same emptiness as an empty one, so the fix is to say that:

-    const itemsCount = items.length
+    // No array at all is the same emptiness as an empty one: the rows go and
+    // the placeholder comes, and nothing below ever reaches into it
+    const itemsCount = items?.length

Nothing else in the loop needed a guard. The array is only touched further down when itemsCount is truthy, and TypeScript narrows it from that alone — two attempts to add a ! were rejected by the linter as unnecessary assertions.

The overload arm

The interesting half is the types. Widening the writable overload from WritableSignal<T[]> to WritableSignal<T[] | EmptyValue> looks like the obvious move and is wrong: a signal is invariant, so WritableSignal<Player[]> does not fit the widened parameter, every existing caller falls through to the readable overload, and its rows silently become read-only. Four existing tests failed with Expected 0 arguments, but got 1 on a row write.

So the nullable array is a separate arm of the same signature:

// A signal is invariant, so the nullable array is a separate arm rather than
// a widening of the first: `WritableSignal<T[]>` does not fit
// `WritableSignal<T[] | EmptyValue>`, and widening would take the rows of
// every existing caller down to read-only
export function for_<T>(
  $items: WritableSignal<T[]> | WritableSignal<T[] | EmptyValue>,
  track?: (item: T, index: number) => unknown
): (
  each_: WritableEach<T>,
  else_?: () => Child
) => Child

The readable and static overloads widen normally.

Tests

One test walking all four transitions: absent on first render shows the placeholder, absent → list shows the row, list → absent brings the placeholder back, and absent → [] leaves it alone rather than tearing it down and building it again. 110 pass.

Size: +1 B gzip on all publics, average usage unchanged, every pin stays where it is.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.32%. Comparing base (42e613b) to head (244f09a).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #205   +/-   ##
=======================================
  Coverage   85.32%   85.32%           
=======================================
  Files         139      139           
  Lines        3142     3142           
  Branches      591      591           
=======================================
  Hits         2681     2681           
  Misses        332      332           
  Partials      129      129           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.