Error for Lint/MixedCaseRange when invalid utf8 and offense exists

[Earlopain]

Expected behavior

Offense gets autocorrected.

Actual behavior

RuboCop throws an error. This is because the range for the regexp literals does not contain the correct range.

Steps to reproduce the problem

Note that – below is the "em dash", not the usual -.

it 'registers an offense for mixed range with invalid byte sequence in UTF-8' do
  expect_offense(<<~'RUBY'.sub(/\#{message}/, message))
    foo = /[\–A-z]/
              ^^^ #{message}
  RUBY

  expect_correction(<<~'RUBY')
    foo = /[\–A-Za-z]/
  RUBY
end

RuboCop version

$ [bundle exec] rubocop -V
1.62.0 (using Parser 3.3.0.4, rubocop-ast 1.31.1, running on ruby 3.3.0) [x86_64-linux]
  - rubocop-performance 1.20.2
  - rubocop-rake 0.6.0
  - rubocop-rspec 2.27.1