[ruby/csv] Fix a bug that the same line is used multiple times

GitHub: fix https://github.com/ruby/csv/pull/279

It's happen when:

* `keep_start`/`keep_{drop,back}` are nested.
  (e.g.: `strip: true, skip_lines: /.../`)
* Row separator is `\r\n`.
* `InputScanner` is used. (Small input doesn't use `InputScanner`)

Reported by Gabriel Nagy. Thanks!!!

https://github.com/ruby/csv/commit/183635ab56

lib/csv/parser.rb

@@ -201,7 +201,8 @@ class CSV
           # trace(__method__, :rescan, start, buffer)
           string = @scanner.string
           if scanner == @scanner
-            keep = string.byteslice(start, string.bytesize - start)
+            keep = string.byteslice(start,
+                                    string.bytesize - @scanner.pos - start)
           else
             keep = string
           end

test/csv/parse/test_skip_lines.rb

@@ -115,4 +115,12 @@ class TestCSVParseSkipLines < Test::Unit::TestCase
                  CSV.parse("a,b\r\n,\r\n",
                            :skip_lines => /^,+$/))
   end
+
+  def test_crlf_strip_no_last_crlf
+    assert_equal([["a"], ["b"]],
+                 CSV.parse("a\r\nb",
+                           row_sep: "\r\n",
+                           skip_lines: /^ *$/,
+                           strip: true))
+  end
 end