Fix nested class scope index

[regularkevvv]

This Pull Request closes #4555.

How to reproduce:

// Class expression constructor with nested class
new (class{constructor(){class D{}}})();

// Static block with nested class
(class{static{class D{}}});

Both crash with: index out of bounds: the len is 0 but the index is 0

Additional patterns that trigger the bug:

// Constructor variants
[new (class{constructor(){class D{}}})()]
({x: new (class{constructor(){class D{}}})()})
new (class{ m(){} constructor(){class D{}} })()
new (class{ static m(){} constructor(){class D{}} })()
new (class{ get g(){return 1} constructor(){class D{}} })()
new (class{ #m(){} constructor(){class D{}} })()
new (class{ ['p']=1; constructor(){class D{}} })()

// Static block variants
(class{ static{let v=1} static{class D{}} })
(class{static{let v=1;((v)=>{class D{}})(v)}})

It changes the following:

• Fixed ScopeIndexVisitor::visit_class_expression_mut to call visit_function_like with force_function_scope: true for constructors, matching visit_class_declaration_mut. Previously it called visit_function_expression_mut, which forwards contains_direct_eval() as the last argument (not an unconditional false). For class constructors we must force function scope regardless of direct eval to match bytecode compilation.
• Fixed ScopeIndexVisitor::visit_class_element_mut for StaticBlock to pass true for force_function_scope. Previously it passed contains_direct_eval, which is correct for escape analysis but not for scope-index layout in ScopeIndexVisitor.
• This change only affects scope-index layout (ScopeIndexVisitor). Escape analysis still uses contains_direct_eval in BindingEscapeAnalyzer, so direct eval-based binding escape behavior is unchanged.

Both class expression constructors and static blocks always have HAS_FUNCTION_SCOPE set during bytecode compilation, so the scope analyzer must account for this when assigning scope indices.

Added two regression tests covering both cases

[github-actions]

Test262 conformance changes

Test result	main count	PR count	difference
Total	52,862	52,862	0
Passed	49,472	49,472	0
Ignored	2,249	2,249	0
Failed	1,141	1,141	0
Panics	0	0	0
Conformance	93.59%	93.59%	0.00%

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 56.82%. Comparing base (6ddc2b4) to head (73cb7bb).
⚠️ Report is 668 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4592      +/-   ##
==========================================
+ Coverage   47.24%   56.82%   +9.57%     
==========================================
  Files         476      548      +72     
  Lines       46892    60008   +13116     
==========================================
+ Hits        22154    34097   +11943     
- Misses      24738    25911    +1173     

☔ 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.

[jedel1043]

The force_function_scope parameter in visit_function_like doesn't exist...

[jedel1043]

Looks good!