Nest sublists in list items under blocksInterruptParagraphs

[dereuromark]

Problem

blocksInterruptParagraphs treated a sublist inconsistently across nesting levels. At the top level a list already interrupts a paragraph under this option, but inside a list item allowsImmediateNestedBlock() explicitly excluded list markers (!$isListMarker), so a blockquote, heading, fence or table nested in an item while a sublist did not. A list is a block, so the two cases should behave the same.

- Item
  - a
  - b

Under blocksInterruptParagraphs this used to fold - a / - b into the item as literal text; only nestedListsWithoutBlankLine would nest it.

Change

The in-item path now delegates list markers to the same startsNewBlockSignificant() decision the top-level paragraph-interruption path uses. Behavior is now identical across nesting levels:

• A real (multi-item) sublist nests inside a list item.
• A lone single-item - x stays literal, via the same lone-marker lookahead that keeps x = 5 / - 3 as prose at the top level.
• Recognized openers match the top level exactly: bullets and 1./1) ordered (the CommonMark guard that keeps 2. apples or (1) note as prose). A 2. or (1) sublist stays literal under interruption alone, the same as at top level.

This makes nestedListsWithoutBlankLine a clear narrower subset: it still nests a lone sublist marker without lookahead, which interruption's lookahead folds in. The deprecated broad nestedBlocksInLists is unchanged.

Notes

The recognized-marker set is intentionally restricted to match top-level interruption, not the canonical parseListItemMarker coverage. Broadening it would make in-item nesting more permissive than top-level interruption and reintroduce the very cross-level inconsistency this fixes; a regression test pins that parity.

Tests cover multi-item nesting, the lone-marker fold, the restricted marker set, and the subset relationship. Full suite green (2487 tests) including the official djot suite; phpstan and phpcs clean.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.87%. Comparing base (1a36893) to head (bfc5edc).

Additional details and impacted files

@@            Coverage Diff            @@
##             master     #209   +/-   ##
=========================================
  Coverage     91.87%   91.87%           
+ Complexity     3489     3488    -1     
=========================================
  Files           104      104           
  Lines          9860     9860           
=========================================
  Hits           9059     9059           
  Misses          801      801           

☔ View full report in Codecov by Sentry.
📢 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.