Refactor code to ignore invalid byte sequences.

The previous implementation made the `classify` method too long for
Rubocop. It also was ignoring invalid byte sequences in only one of the
two places where we loop through the lines to look for :nocov:.

lib/simplecov/lines_classifier.rb

@@ -16,21 +16,30 @@ def self.no_cov_line
       @no_cov_line ||= /^(\s*)#(\s*)(\:#{SimpleCov.nocov_token}\:)/
     end
 
+    def self.no_cov_line?(line)
+      line =~ no_cov_line
+    rescue ArgumentError
+      # E.g., line contains an invalid byte sequence in UTF-8
+      false
+    end
+
+    def self.whitespace_line?(line)
+      line =~ WHITESPACE_OR_COMMENT_LINE
+    rescue ArgumentError
+      # E.g., line contains an invalid byte sequence in UTF-8
+      false
+    end
+
     def classify(lines)
       skipping = false
 
       lines.map do |line|
-        begin
-          if line =~ self.class.no_cov_line
-            skipping = !skipping
-            NOT_RELEVANT
-          elsif skipping || line =~ WHITESPACE_OR_COMMENT_LINE
-            NOT_RELEVANT
-          else
-            RELEVANT
-          end
-        rescue ArgumentError
-          # E.g., line contains an invalid byte sequence in UTF-8
+        if self.class.no_cov_line?(line)
+          skipping = !skipping
+          NOT_RELEVANT
+        elsif skipping || self.class.whitespace_line?(line)
+          NOT_RELEVANT
+        else
           RELEVANT
         end
       end

lib/simplecov/source_file.rb

@@ -182,7 +182,7 @@ def process_skipped_lines(lines)
       skipping = false
 
       lines.each do |line|
-        if line.src =~ SimpleCov::LinesClassifier.no_cov_line
+        if SimpleCov::LinesClassifier.no_cov_line?(line.src)
           skipping = !skipping
           line.skipped!
         elsif skipping