…ation

Adds the concept of a "handler" function to UT assert.  A handler
is basically the custom logic that exists between the hook function
and the return to the stub caller.  In current UT stubs, this is
hard coded, and it generally comprises setting output parameters
and translating return values as needed.

This concept adds the basic plumbing to allow the handler to be
configured just like a hook function already does.  The difference
is that the handler is directly responsible for setting all outputs.

This also includes a script to auto-generate stub functions that
match this pattern.  Given an API header file, the script extracts
the declarations, and generates a source file with stub definitions
that rely on a separate handler to deal with the needed outputs.

Note this initial commit only adds the basic framework.  It does
not change any existing stubs or tests, and is fully backward
compatible, as it is a new feature and it is a no-op unless
actually configured/used by the stub or test case.  Follow on
commits will update the stubs to use this pattern.

To imporove scriptability of generating the shared layer stubs, makes
two minor adjustments:

- adds a typedef for the callback in OS_ObjectIdIteratorProcessEntry.  This
  avoids the complex syntax of function pointer argument, which is hard to
  read but also harder to parse in a script, too.
- Splits the "os-shared-printf.h" header into two parts, adding a new
  header file "os-shared-console.h".  This is because OS_ConsoleOutput_Impl
  is implemented in a separate unit in the source, and this allows a better
  relationship between stub files and header files.

The "built-in" logic from existing stubs is converted to a
handler function and moved to a different source file.  The
generated stub references this function if it exists and
uses it as a default handler, so the default behavior is not
changed.

In this pattern the stubs in this directory are strictly
limited to the same public APIs that are defined in header
files under src/os/inc.  This has the side effect of removing
some internal stubs required for coverage test.  Those stubs
will be reinstated in the coverage specific stubs where other
internal functions are.

Update the coverage-specific (shared layer internal) stubs to use
generated stubs.

The OS_SocketAccept call was using the incorrect token, using
the server/acceptor socket when it should have used the connection
socket.

The bug overwrote data in the acceptor socket, but it would only cause
an issue when the user attempted to use the server socket to accept
a second connection, but the tests only performed a single connection.

This also improves the network-api-test to do multiple connections,
re-using the same acceptor socket between them.

Update the API documentation for OS_read/write/TimedRead/TimedWrite.
Clarify that zero will be returned on EOF condition, and in the
case of the timed API calls, the OS_ERR_TIMEOUT status code will
be returned if the timeout expired without the handle becoming
readable/writable during that time.

This is intended to allow the caller to differentiate between a
handle which is in a state where it cannot read/write data (e.g.
at EOF, or a pipe/socket with remote end closed) and handle which
is simply idle or busy.

The stream test was sending an 8-bit count pattern to validate
the connection, but was not using the provided API to do so.

This provides the exact same test but uses the provided API
from UtAssert to implement it, rather than duplicating the logic.

Notably, this also confirms that the tool functions work as
expected, since there were no other OSAL tests using this API.

Fix #970, update documentation for read/write

Fix #419 #729 #954, Adds local makefile and bundle/local unit test actions with coverage verification

Fix #969, socket accept using incorrect record