REXML is dead. Long live Nokogiri.

Modify Maruku to use Nokogiri instead of REXML.
Produces a 3-fold speedup in the #to_html method.

vendor/plugins/maruku/spec/block_docs/paragraph_rules/dont_merge_ref.md

@@ -0,0 +1,42 @@
+Paragraphs eats everything, but not link definitions.
+*** Parameters: ***
+{}
+*** Markdown input: ***
+Paragraph
+[google1]: #
+
+Paragraph
+ [google2]: #
+
+Paragraph
+  [google3]: #
+
+*** Output of inspect ***
+md_el(:document,[
+	md_par(["Paragraph"]),
+	md_ref_def("google1", "#", {:title=>nil}),
+	md_par(["Paragraph"]),
+	md_ref_def("google2", "#", {:title=>nil}),
+	md_par(["Paragraph"]),
+	md_ref_def("google3", "#", {:title=>nil})
+],{},[])
+*** Output of to_html ***
+<p>Paragraph</p>
+
+<p>Paragraph</p>
+
+<p>Paragraph</p>
+*** Output of to_latex ***
+Paragraph
+
+Paragraph
+
+Paragraph
+*** Output of to_md ***
+Paragraph
+
+Paragraph
+
+Paragraph
+*** Output of to_s ***
+ParagraphParagraphParagraph

vendor/plugins/maruku/spec/block_docs/paragraph_rules/tab_is_blank.md

@@ -0,0 +1,24 @@
+Paragraphs eat blank lines.
+The following are two paragraphs:
+*** Parameters: ***
+{}
+*** Markdown input: ***
+Paragraph1
+	
+Paragraph2
+*** Output of inspect ***
+md_el(:document,[md_par(["Paragraph1"]), md_par(["Paragraph2"])],{},[])
+*** Output of to_html ***
+<p>Paragraph1</p>
+
+<p>Paragraph2</p>
+*** Output of to_latex ***
+Paragraph1
+
+Paragraph2
+*** Output of to_md ***
+Paragraph1
+
+Paragraph2
+*** Output of to_s ***
+Paragraph1Paragraph2