Filters

- Fixed a naming conflict preventing `ByteString` and `Unicode` from being chained.
- `BaseFilter.normalize` has been renamed to `resolve_filter`.

- Made readthedocs the official homepage for this project.

- It is now possible to provide a `decimal.Decimal` value for the `max_precision` parameter to `Decimal.__init__`.
- Made `BaseFilter._normalize` a public method.
- Internal reorganization, refactored a bunch of "circular import" kludges.
- Added lots of documentation.
- Cleaned up comments and docstrings for reStructuredText.

- `skip_value_check` can now be specified as a class-level attribute in test cases if you want to disable filtered value comparisons across all tests in that test case.
- Previously, `skip_value_check` had to be included in _every_ call to `assertFilterPasses` and `assertFilterErrors.
- To use this feature, set `skip_value_check=True` in the test case definition. E.g.:
`
class SomeTestCase(TestCase):
skip_value_check = True
`

- Implemented `Array` filter (similar to `Type(Sequence)`, but rejects string values).
- Added `typing.Mapping` and `typing.Sequence` to `JSON_ALIASES`.

- Minor code cleanup.
- Fixed an unexpected error when a filter test fails.