Gracefully handle encoding mishaps

We are getting reports like in #14 where we're not selecting the best
encoding for a file in some cases. While this is usually user error, we
can try to fix things by attempting to change the encoding to UTF-8 as
that's usually what is intended.

Piggy-back on the Temple library's `Encoding` filter to clean this up
for us.

lib/slim_lint/document.rb

@@ -37,13 +37,22 @@ def initialize(source, options)
     # @param source [String] Slim code to parse
     # @raise [SlimLint::Exceptions::ParseError] if there was a problem parsing the document
     def process_source(source)
+      @source = process_encoding(source)
       @source = strip_frontmatter(source)
       @source_lines = @source.split("\n")
 
       engine = SlimLint::Engine.new(file: @file)
       @sexp = engine.parse(source)
     end
 
+    # Ensure the string's encoding is valid.
+    #
+    # @param source [String]
+    # @return [String] source encoded in a valid encoding
+    def process_encoding(source)
+      ::Temple::Filters::Encoding.new.call(source)
+    end
+
     # Removes YAML frontmatter
     def strip_frontmatter(source)
       if config['skip_frontmatter'] &&