This is probably a better approach than potentially having a pointer
that points to garbage on the stack and gets written to by cJSON.

Add a pointer to an end position of parsing to the cJSON_Configuration
object. (Essentially like return_parse_end, but as offset instead of
pointer).

Also introduces a allow_data_after_json property in the internal
configuration.

Check the size to prevent comparing objects equal if they are prefixes
of each other.

This should improve performance

Thanks @Tangerino for suggesting this optimisation.

This is so that NaN and INFINITY values can be produced.

Release Version 1.7.1

Thanks @liuyunbin for reporting this in #230

Thanks @liuyunbin for reporting this in #230

Thanks @liuyunbin for reporting this in #230

Release Version 1.7.0

New documentation