Tests were not correctly testing the situations without a table close.
Fixed tests and then fixed tokenizers for failing tests. Also refactored
pytokenizer to more closely match the ctokenizer by only holding the
`_parse` methods in the try blocks and no other code.

One-line fix

Python 3.4 compiles C extensions with the
`-Werror=declaration-after-statement` flag that enforces C90 more
strictly than previous versions. Move all statements after declarations
to make sure this extension builds on 3.4.

For wiki syntax tables, add `wiki_style_separator` as an attribute
for the Tag node. Also reorder `closing_wiki_markup` property and tests
to match its place in the constructor.

Table tags no longer self-closing. Rows and cells now contain their
contents. Also refactored out an `emit_table_tag` method.
Note: this will require changes to the Tag node and possibly the builder,
those changes will be in the next commit.

Self-closing wiki syntax tags have incorrectly ordered wiki syntax
and padding, fixed the ordering.

CTokenizer is completely implemented in this commit - it didn't
make much sense to me to split it up. All tests passing, memory test
shows no leaks on Linux.

For the C tokenizer, include `<stdint.h>` and use `uint64_t` instead
of `int` for context. Changes to tables mean that context can be
larger than 32 bits, and it is possible for `int` to only have 16
bits anyways (though this is very unlikely).

For C compatability, switch table cell end to return the stack.
Now context is kept by using `keep_context` when calling `self._pop()`.

Various changes to avoid returning tuples - working on the C tokenizer
made me realize this was a bad idea for compatability/similarity between
the two.

Fix problem in which invalid table attributes were being parsed
incorrectly. Added tests.

Fix problem in which fake table closes were causing a problem inside
cells. Changed inline table handling to fix this.

Padding now included on all wiki table cells. With wiki table cells
that include attributes, `wiki_markup` is also included (unchanged).

Make sure py tokenizer methods only call methods that have been declared
earlier. Not necessary but makes it much easier to maintain/write
the C tokenizer if methods are in the same order.

Removed the `StopIteration()` exception for handling table style
and instead call `_handle_table_cell_end()` with a new parameter.
Also added some random tests for table openings.

Changed row recursion handling to make sure the tag is emitted even
when hitting recursion limits. Need to test table recursion to make
sure that works. Also fixed a bug in which tables were eating the
trailing token. Added several tests for rows and trailing tokens with
tables.

Tables and rows use newlines as padding, partly because these characters
are pretty important to the integrity of the table. They might need
to be in the preceding whitespace of inner tags instead as padding after,
not sure.

Added comments, tried to keep to 80 character lines.

Added support for allowing different wiki syntax for replacing the opening
and closing tags. Added for table support.

Started styling attributes for table row and table start. Still not entirely
sure about this, definitely need to make changes regarding padding.

Support for header cells was mostly in already, just needed minor
changes. Added two tests as well.

Started support for parsing table style attributes. I suspect some
of this is incorrect, need to add more tests to see.

Added another stack layer for tokenizing table cells because of
styling/correctness of implementation. Added many tests cases.

Started parsing table support and added the start of table support.
This is a big commit (ugh) and it should probably be split up into
multiple smaller ones if possible, but that seems unworkable as of
right now because of all the dependencies. Also breaks tests of
CTokenizer (double ugh) because I haven't started table support there.

May want to pick line by line on this commit later but I need to save
my work for now.

* Parameters with non-integer keys can no longer be created with
  showkey=False, nor have the value of this attribute be set to False
  later.
* Calling Template.remove() with a Parameter object that is not part of
  the template now raises ValueError instead of doing nothing.
* Added tests for HTMLEntity._unichr() being called with out-of-range
  codepoints.
* Added tests for Tag.__children__() and Tag.__showtree__() involving
  attributes that have no values.

Before:
$ python -m timeit -s 'from mwparserfromhell.parser.tokens import Text' 'Text(text = "abc")'
1000000 loops, best of 3: 0.91 usec per loop

After:
$ python -m timeit -s 'from mwparserfromhell.parser.tokens import Text' 'Text(text = "abc")'
1000000 loops, best of 3: 0.223 usec per loop

Patch mostly by @spointy.