Py-kim

-----------------------
Fix issue where min/max would not work on Decimal fields when passed as string

-----------------------

Fix issue where field.opts.attribute_name would always be None

-----------------------

This release contains a number of improvements and changes to existing field options as
well as some new Field types

https://github.com/mikeywaites/kim/milestone/4?closed=1

* New Float field (Margaferrez)
* Min/Max options for Decimal field (Margaferrez)
* String Min/Max validation options (Margaferrez)
* Add blank option to String to disallow empty string
* make field default apply when serializing and marshaling
* Error now raise when trying to serialize None
* Default String field to unicode type

-----------------------
Fix version of iso8601 dependency which was broken by the latest release of that project

-----------------------

This release focussed solely on improving the peformance of Kim during serialization cycles. Overall Kim
is now around 75% faster.

* Deprecated `Pipe` class. This was causing a huge amount of overhead in `Pipelines`
* Reduced the foot print of `Pipeline` and `Session` with usage of __slots__
* removed marshal_extra_X and serialize_extra_X. Additional pipes are now added to the Pipeline at compile time.
* Avoid duplicate calls to the _MapperRegistry
* optimised the number of time get_mapper_session is called especially in field.Collection runs.
* Added benchmark suite to repo which is run during circle.ci builds

-----------------------

Removal of Try/Except block from update_output_to_name. This yielded ~10% performance increase.