Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@crates/biome_js_analyze/src/lint/nursery/no_implied_eval.rs`:
- Around line 80-83: The mapping claims coverage of ESLint's "no-new-func" but
this rule (in no_implied_eval.rs) only inspects
setTimeout/setInterval/setImmediate calls, so either remove
RuleSource::Eslint("no-new-func").inspired() from the list or extend the lint to
detect Function constructor usage; to fix, either delete the
RuleSource::Eslint("no-new-func").inspired() entry in the rule's sources array,
or add AST checks in the no_implied_eval lint for NewExpression and
CallExpression nodes whose callee is an Identifier named "Function" (and also
handle possible MemberExpression cases if needed) and emit the same diagnostic
so the rule truly covers Function(...) / new Function(...).

---

Nitpick comments:
In @.changeset/tidy-shirts-cut.md:
- Line 5: Update the changeset description so it speaks to end users: replace
the maintainer-facing metadata mention of "noImpliedEval" and "no-new-func as a
rule source" with a user-facing sentence explaining the change (e.g., "Add
no-new-func as a source for noImpliedEval rule to improve detection of new
Function() and eval-like usage") and add the recommended: true flag to the
changeset metadata so the release notes clearly state this rule is recommended.

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@crates/biome_js_analyze/src/lint/nursery/no_implied_eval.rs`:
- Around line 107-130: The string-argument branch currently runs for both
JsNewExpression and JsCallExpression (because Query is JsNewOrCallExpression)
causing false positives for constructs like new setTimeout("..."); restrict that
branch to call expressions only: after resolving node (in run) verify the node
is a call expression (e.g., via node.as_call_expression() /
node.is_call_expression() or pattern-match the JsNewOrCallExpression into a
JsCallExpression) before checking args and calling is_string_argument, and only
return Some(NoImpliedEvalState::StringArgument) when the node is a call
expression.

In `@crates/biome_js_analyze/tests/specs/nursery/noImpliedEval/valid.js`:
- Around line 68-79: The file has two top-level declarations named Function (a
class at the top and a later function), causing a syntax error; split them into
separate scopes so both shadowing cases are valid—e.g., wrap one of the
top-level cases in its own block or IIFE (or move the later function into the
existing nestedClass or another function scope) so the class declaration `class
Function {}` and the function declaration `function Function() {}` are not in
the same top-level scope; ensure the `new Function()` calls remain in scope that
sees the intended shadowed binding so the rule tests still exercise both
shadowing scenarios.

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@crates/biome_js_analyze/src/lint/nursery/no_implied_eval.rs`:
- Around line 218-227: The .clone() on expression is unnecessary; in
is_global_function_constructor call global_identifier with the existing
reference rather than cloning—replace global_identifier(&expression.clone())
with global_identifier(expression) (referencing the function
is_global_function_constructor, the parameter expression, and the
global_identifier call).