API change: This adds a new exported function in cmark.h.

Closes #330.

Use zero-terminated C strings and a separate length field instead of
cmark_chunks. Literal inline text will now be copied from the parent
block's content buffer, slowing the benchmark down by 10-15%.

The node struct never references memory of other nodes now, fixing #309.
Node accessors don't have to check for delayed creation of C strings,
so parsing and iterating all literals using the public API should
actually be faster than before.

Use zero-terminated C strings instead of cmark_chunks without storing
the length. This introduces a few additional strlen computations,
but overhead should be low.

Allows to reduce size of struct cmark_node later.

Removes CMARK_OPT_SAFE from options.

Adds CMARK_OPT_UNSAFE, with the opposite meaning.
The new default behavior is to suppress raw HTML and
potentially dangerous links.  The CMARK_OPT_UNSAFE
option has to be set explicitly to prevent this.

--------------------------------------------------------
NOTE: This change will require modifications in
bindings for cmark and in most libraries and programs
that use cmark.
--------------------------------------------------------

Closes #239, #273.

Borrows heavily from @kivikakk's patch in github/cmark-gfm#123.

With current HTML escaping, sometimes we may produce an XML tag like

    <code_block>  </code_block>

Many XML parsers consider these spaces insignificant and strip them
out but we need this. There's actually a test case like this in
spec.txt (search "A code block can have all empty lines as its
content:") and without proper hinting, an external xml->html converter
will fail the spec.

XML standard covers this case. If xml:space is "preserve", then
whitespaces are significant and should be kept. Add this hint for
text, code, code_block, html_inline and html_block tags.