ast.rb: Fix bug for source of multibyte characters

[alpaca-tc]

RubyVM::AbstractSyntaxTree::Node's #first_column and #last_column return byte positions, but existing implementations did not consider multi-byte.

ast = RubyVM::AbstractSyntaxTree.parse(%{a("§");b("©")}, keep_script_lines: true)
a_fcall, b_fcall = ast.children[2].children

puts a_fcall.source
puts b_fcall.source

# before
a("§");
("©")

# after
a("§")
b("©")