No-op blank line

Compilers generally provide a much smarter strlen than ours.

Writer EndObject missing-value fix

Null handler fix

We should not malloc the null-handler object and cast to OutputHandler;
we need to actually invoke the constructor via placement new.

Add unit test for Issue 848 (segfault in ~Document)

PrettyWriter EndObject will now also assert if called when a key is
missing its matching value.

Documented existing assertions in EndObject
Added new assertion in EndObject to catch error condition where objects
are ended with a key but no matching value.

Add unit tests expecting an assertion when writing an object with a key
but no value.

Fix creating the nuget package with Raggles' fork of CoApp

Token-by-token pull parsing

Missing "internal" namespace for StrLen

include/rapidjson/pointer.h:243:40: error: 'StrLen' was not declared in this scope
         return Append(name, StrLen(name), allocator);

NaN/Inf parse fix

A failed half-consume of “NaN” now returns “value invalid” instead of
attempting to consume an “Inf”.

Noticed that the reader could over-consume “NaN” if token terminated in
the middle.

Add more tests! Good for coverage.

Runs about 1-2% faster (original speed) by running in a tight loop, at
the expense of slight code duplication with IterativeParseNext.

Improved reporting

std::to_string can’t be used because it requires C++11.

Example code to demonstrate how the token-pulling reader can be used.

Doesn’t seem to affect timings in perftest on my machine, but it may
help others.

New unit test which ensures that IterativeParseNext always generates
exactly one element at a time, and that calling IterativeParseNext on a
complete document is harmless and generates zero events.

Further improvement to perftest and hoping to make coveralls happy.

Slight performance improvement over previous submission