Symbol encoding handling

[lopopolo]

String#intern and Symbol#to_s have some nuances around encoding that are not properly modeled by Artichoke:

• String#intern raises an EncodingError if the string is not valid per its encoding, but maybe only for multibyte encodings:

  [3.1.2] > a = "abc\xFF"
  => "abc\xFF"
  [3.1.2] > a.encoding
  => #<Encoding:UTF-8>
  [3.1.2] > a.valid_encoding?
  => false
  [3.1.2] > a.intern
  (irb):17:in `intern': invalid symbol in encoding UTF-8 :"abc\xFF" (EncodingError)
          from (irb):17:in `<main>'
          from /usr/local/var/rbenv/versions/3.1.2/lib/ruby/gems/3.1.0/gems/irb-1.4.1/exe/irb:11:in `<top (required)>'
          from /usr/local/var/rbenv/versions/3.1.2/bin/irb:25:in `load'
          from /usr/local/var/rbenv/versions/3.1.2/bin/irb:25:in `<main>'
  [3.1.2] > b = "abc\xFF".force_encoding(Encoding::ASCII)
  => "abc\xFF"
  [3.1.2] > b.encoding
  => #<Encoding:US-ASCII>
  [3.1.2] > b.valid_encoding?
  => false
  [3.1.2] > b.intern
  => :"abc\xFF"
  

• Symbol#to_s is influenced by the encoding of the string that was interned:

  [3.1.2] > b = "abc\xFF".force_encoding(Encoding::ASCII)
  => "abc\xFF"
  [3.1.2] > b.encoding
  => #<Encoding:US-ASCII>
  [3.1.2] > b.valid_encoding?
  => false
  [3.1.2] > t = b.intern
  => :"abc\xFF"
  [3.1.2] > t.to_s.encoding
  => #<Encoding:US-ASCII>
  [3.1.2] > t.to_s.valid_encoding?
  => false
  [3.1.2] > u = c.intern
  => :"abc 😄"
  [3.1.2] > u.to_s.encoding
  => #<Encoding:UTF-8>
  [3.1.2] > u.to_s.valid_encoding?
  => true
  [3.1.2] > d = "cvb\xFF".b
  => "cvb\xFF"
  [3.1.2] > d.encoding
  => #<Encoding:ASCII-8BIT>
  [3.1.2] > d.valid_encoding?
  => true
  [3.1.2] > v = d.intern
  => :"cvb\xFF"
  [3.1.2] > v.to_s.encoding
  => #<Encoding:ASCII-8BIT>
  [3.1.2] > e = "bnm\xFF".force_encoding(Encoding::ASCII)
  => "bnm\xFF"
  [3.1.2] > e.encoding
  => #<Encoding:US-ASCII>
  [3.1.2] > e.valid_encoding?
  => false
  [3.1.2] > w = e.intern
  => :"bnm\xFF"
  [3.1.2] > w.to_s.encoding
  => #<Encoding:US-ASCII>
  [3.1.2] > w.to_s.valid_encoding?
  => false