in relation to solving issue #784, this commit enables the schema to recognise the "default" property, and avoids a missing property error when a default is given in the schema

Prevent int underflow when parsing exponents

When parsing negative exponents, the current implementation takes
precautions for |exp| to not underflow int.

But that is not sufficient: later on [1], |exp + expFrac| is also
stored to an int - so we must ensure that the sum stays within int
representable values.

Update the exp clamping logic to take expFrac into account.

[1] https://github.com/Tencent/rapidjson/blob/master/include/rapidjson/reader.h#L1690

Added const for Reader methods

Fix compilation on windows with clang

Use rvalue refs with clang-cl

Allow user configuration of default chunk capacity.

Added macro RAPIDJSON_ALLOCATOR_DEFAULT_CHUNK_CAPACITY to allow default chunk capacity to be lowered for embedded devices with < 64k stack sizes

add remote ref pointer to schemaMap_

Ensure a preprocessor directive is only applied to MSVC.

Adding a few missing includes in the doc

Change long/ulong as int/uint on MSC unit tests to be more inline wit…

Change long/ulong as int/uint on MSC unit tests to be more inline with other templated functions unit tests

Fixes #1198

Support long and unsined long as int and unsigned on Microsft platforms

Fix warnings: dereference of null pointer

Ignore GCC 8 warnings.

Misc. typos

Found via `codespell -q 3` in  downstream https://github.com/BlueBrain/Brayns

This reverts commit 0d2580f1.

Fix #1014

Add transcoding/validation to Writer::RawValue()

Fix #1152

Fix #1174