Assume an eval context for Prism::Translation::Parser

This is similar to davidwessman/syntax_tree-erb#81 but for RuboCop
The parser gem doesn't support these types of checks,
see https://github.com/whitequark/parser?tab=readme-ov-file#syntax-check-of-block-exits

While this is technically a bug in the parser gem, it does increase compatibility
and allows prism to be used when linting erb or haml with a RuboCop extension.

lib/prism/translation/parser.rb

@@ -46,7 +46,7 @@ def parse(source_buffer)
         source = source_buffer.source
 
         offset_cache = build_offset_cache(source)
-        result = unwrap(Prism.parse(source, filepath: source_buffer.name, version: convert_for_prism(version)), offset_cache)
+        result = unwrap(Prism.parse(source, filepath: source_buffer.name, version: convert_for_prism(version), scopes: [[]]), offset_cache)
 
         build_ast(result.value, offset_cache)
       ensure
@@ -59,7 +59,7 @@ def parse_with_comments(source_buffer)
         source = source_buffer.source
 
         offset_cache = build_offset_cache(source)
-        result = unwrap(Prism.parse(source, filepath: source_buffer.name, version: convert_for_prism(version)), offset_cache)
+        result = unwrap(Prism.parse(source, filepath: source_buffer.name, version: convert_for_prism(version), scopes: [[]]), offset_cache)
 
         [
           build_ast(result.value, offset_cache),
@@ -78,7 +78,7 @@ def tokenize(source_buffer, recover = false)
         offset_cache = build_offset_cache(source)
         result =
           begin
-            unwrap(Prism.parse_lex(source, filepath: source_buffer.name, version: convert_for_prism(version)), offset_cache)
+            unwrap(Prism.parse_lex(source, filepath: source_buffer.name, version: convert_for_prism(version), scopes: [[]]), offset_cache)
           rescue ::Parser::SyntaxError
             raise if !recover
           end