Fixup benchmark and speedup a little, fixes #141

[ojab]

With README.md from this repo:
1ee03a8 Fix benchmark, switch it to benchmark-ips and add xml + Node#to_html

input size = 9206 bytes

Warming up --------------------------------------
           redcarpet     1.998k i/100ms
commonmarker with to_html
                       652.000  i/100ms
commonmarker with to_xml
                       651.000  i/100ms
commonmarker with ruby HtmlRenderer
                       140.000  i/100ms
commonmarker with render_doc.to_html
                         1.021k i/100ms
            kramdown    17.000  i/100ms
Calculating -------------------------------------
           redcarpet     21.305k (± 1.5%) i/s -    107.892k in   5.065176s
commonmarker with to_html
                          6.127k (± 3.9%) i/s -     30.644k in   5.009868s
commonmarker with to_xml
                          5.775k (± 3.9%) i/s -     29.295k in   5.080767s
commonmarker with ruby HtmlRenderer
                          1.278k (± 6.1%) i/s -      6.440k in   5.057872s
commonmarker with render_doc.to_html
                          9.606k (±11.4%) i/s -     47.987k in   5.052644s
            kramdown    154.707  (± 5.8%) i/s -    782.000  in   5.075266s

Comparison:
           redcarpet:    21305.5 i/s
commonmarker with render_doc.to_html:     9606.0 i/s - 2.22x  (± 0.00) slower
commonmarker with to_html:     6126.7 i/s - 3.48x  (± 0.00) slower
commonmarker with to_xml:     5774.8 i/s - 3.69x  (± 0.00) slower
commonmarker with ruby HtmlRenderer:     1278.3 i/s - 16.67x  (± 0.00) slower
            kramdown:      154.7 i/s - 137.72x  (± 0.00) slower

272a721 Use default allocator for parser and rewrite creation a bit

input size = 9206 bytes

Warming up --------------------------------------
           redcarpet     1.915k i/100ms
commonmarker with to_html
                         1.197k i/100ms
commonmarker with to_xml
                         1.213k i/100ms
commonmarker with ruby HtmlRenderer
                       131.000  i/100ms
commonmarker with render_doc.to_html
                       989.000  i/100ms
            kramdown    16.000  i/100ms
Calculating -------------------------------------
           redcarpet     20.021k (± 1.7%) i/s -    101.495k in   5.070895s
commonmarker with to_html
                         11.790k (± 7.5%) i/s -     58.653k in   5.007325s
commonmarker with to_xml
                         11.386k (± 5.9%) i/s -     57.011k in   5.027840s
commonmarker with ruby HtmlRenderer
                          1.254k (± 5.8%) i/s -      6.288k in   5.036073s
commonmarker with render_doc.to_html
                          9.469k (±11.9%) i/s -     47.472k in   5.078444s
            kramdown    138.965  (±10.1%) i/s -    688.000  in   5.006044s

Comparison:
           redcarpet:    20021.4 i/s
commonmarker with to_html:    11789.7 i/s - 1.70x  (± 0.00) slower
commonmarker with to_xml:    11385.7 i/s - 1.76x  (± 0.00) slower
commonmarker with render_doc.to_html:     9468.7 i/s - 2.11x  (± 0.00) slower
commonmarker with ruby HtmlRenderer:     1254.0 i/s - 15.97x  (± 0.00) slower
            kramdown:      139.0 i/s - 144.08x  (± 0.00) slower

936e91017f65aaade3d5894d02623d1a1f816347 Rework rb_markdown_to_html

input size = 9206 bytes

Warming up --------------------------------------
           redcarpet     1.727k i/100ms
commonmarker with to_html
                         1.010k i/100ms
commonmarker with to_xml
                         1.050k i/100ms
commonmarker with ruby HtmlRenderer
                       115.000  i/100ms
commonmarker with render_doc.to_html
                       760.000  i/100ms
            kramdown    12.000  i/100ms
Calculating -------------------------------------
           redcarpet     16.851k (± 9.5%) i/s -     84.623k in   5.078071s
commonmarker with to_html
                         10.699k (± 4.0%) i/s -     53.530k in   5.011280s
commonmarker with to_xml
                         10.413k (± 4.0%) i/s -     52.500k in   5.050548s
commonmarker with ruby HtmlRenderer
                          1.134k (± 2.8%) i/s -      5.750k in   5.075396s
commonmarker with render_doc.to_html
                          8.896k (±11.2%) i/s -     44.080k in   5.023672s
            kramdown    145.532  (± 5.5%) i/s -    732.000  in   5.047857s

Comparison:
           redcarpet:    16851.2 i/s
commonmarker with to_html:    10699.4 i/s - 1.57x  (± 0.00) slower
commonmarker with to_xml:    10413.0 i/s - 1.62x  (± 0.00) slower
commonmarker with render_doc.to_html:     8896.2 i/s - 1.89x  (± 0.00) slower
commonmarker with ruby HtmlRenderer:     1133.9 i/s - 14.86x  (± 0.00) slower
            kramdown:      145.5 i/s - 115.79x  (± 0.00) slower

67f65d08057fd58be844a184043d586d3e8167a5 Rework rb_markdown_to_xml

input size = 9206 bytes

Warming up --------------------------------------
           redcarpet     1.714k i/100ms
commonmarker with to_html
                         1.018k i/100ms
commonmarker with to_xml
                         1.034k i/100ms
commonmarker with ruby HtmlRenderer
                       112.000  i/100ms
commonmarker with render_doc.to_html
                       750.000  i/100ms
            kramdown    14.000  i/100ms
Calculating -------------------------------------
           redcarpet     18.833k (± 2.9%) i/s -     94.270k in   5.010007s
commonmarker with to_html
                         11.402k (± 3.1%) i/s -     57.008k in   5.004576s
commonmarker with to_xml
                         11.328k (± 1.7%) i/s -     56.870k in   5.021896s
commonmarker with ruby HtmlRenderer
                          1.228k (± 2.7%) i/s -      6.160k in   5.020745s
commonmarker with render_doc.to_html
                          9.053k (±11.1%) i/s -     45.000k in   5.039098s
            kramdown    137.681  (±17.4%) i/s -    658.000  in   5.055183s

Comparison:
           redcarpet:    18832.8 i/s
commonmarker with to_html:    11402.2 i/s - 1.65x  (± 0.00) slower
commonmarker with to_xml:    11327.9 i/s - 1.66x  (± 0.00) slower
commonmarker with render_doc.to_html:     9052.8 i/s - 2.08x  (± 0.00) slower
commonmarker with ruby HtmlRenderer:     1227.8 i/s - 15.34x  (± 0.00) slower
            kramdown:      137.7 i/s - 136.79x  (± 0.00) slower

[ojab]

rubocop failures are unrelated, sending to review.

[ojab]

Fixed rubocop, hopefully you don't mind requiring mfa for rubygems.

[gjtorikian]

This all looks really great, thank you. As with any C change I want to take a deeper look at it (beyond the superficial one I've just given), but I don't see anything too outrageous. Thank you for your work. 😃

[kivikakk]

I've given it a once-over and local test; looking good at first blush! ❤️