Doc: Migrate Put special stuff in NuttX header files

Migrate the page "Why can't I put my special stuff in NuttX header
files?" page (https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=139629466) to the official wiki.

Signed-off-by: Ludovic Vanasse <ludovicvanasse@gmail.com>

Documentation/guides/index.rst

@@ -41,3 +41,4 @@ Guides
   armv7m_runtimestackcheck.rst
   disabling_stackdumpdebug.rst
   include_files_board_h.rst
+  specialstuff_in_nuttxheaderfiles.rst

Documentation/guides/specialstuff_in_nuttxheaderfiles.rst

@@ -0,0 +1,72 @@
+=======================================================
+Why can't I put my special stuff in NuttX header files?
+=======================================================
+
+.. warning::
+    Migrated from: https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=139629466
+
+The Problem
+===========
+
+I am very picky about what goes into NuttX header files. I don't accept 
+non-standardized changes to go into them just to permit external code to 
+compile; nor do I accept changes that imply something is implemented in 
+NuttX when it is not (although there are cases like that in the header 
+files now).
+
+* But I want to use `newlib` logic that depends on certain definitions on the 
+  Nuttx header files!
+* But I am trying to compile an application that depends on non-standard 
+  declarations in header files! Or prototypes for functions that are not 
+  provided by NuttX! Or types that are not used by NuttX!
+
+You will find that I am very stubborn on this subject and you will be wasting 
+your time and energy if you try to get kruft included into NuttX header files 
+for your personal purposes.
+
+A Work-Around
+=============
+
+But there is a work-around for my pickiness and stubborn-ness (at least for 
+compilers like GCC that support the GNU extensions). Let's suppose you wanted 
+to add this definition:
+
+.. code-block:: c
+
+    #define I_AM_A_NERD true
+
+to the standard ``time.h`` header file. You submitted a patch to do this and 
+I refused it. Now what?
+
+While I refuse to put non-standard or useless stuff in NuttX header files, 
+there are ways to work around this. Suppose that you create a directory 
+called ``myincludes/`` and in your ``myincludes/`` directory is a header called 
+``time.h``. This ``time.h`` header file consists of:
+
+.. code-block:: C
+
+    #define I_AM_A_NERD true
+    #include_next <time.h>
+
+Then in your ``CFLAGS``, you use an ``-isystem`` setting to include header 
+files from ``myincludes/`` before any header files from the NuttX ``include/`` 
+directory. Then when your application includes ``time.h``, the version of 
+``time.h`` in ``myincludes/`` is the one that will be included. That version 
+will define ``I_AM_A_NERD`` as you want and then include the next file named 
+``time.h`` in the compiler's include path. That file will be the standard 
+``time.h`` header file that is provided in the NuttX ``include/`` directory
+
+In this way you an append or modify any of the NuttX header files to suit 
+your own purposes without my having to accept changes that I do not want 
+into the NuttX repository.
+
+When Does It Make Sense?
+========================
+
+When does it make sense to add new definitions, types, and function prototypes 
+to the NuttX header files? Only under the following conditions:
+
+* The changes are standard and specified in OpenGroup.org
+* The changes are provided by a patch that includes the full, verified 
+  implementation of the feature that uses the types and implements the 
+  functions.