Improves the distinction between PDU data being actively interpreted
or created during the PDU receive or transmit process, and the encoded
form of that data.

CF formerly treated the two as the same, directly referencing the
encoded form of the data.  This creates many repeated translations.
Furthermore, it would sometimes write a modified value back to the
packet in a partially-decoded form, so it was never clear what
was in a given buffer at a given time (it could be native byte
order or network byte order, in the same fields).

This introduces a "logical" buffer which correlates to the CFDP
buffer, but is used for all in-work or temporary value storage.
All values in the logical buffer are normalized to the native
machine, that is they are aligned properly and always in the
correct byte order for the host, so they can be used as normal
values without any need for translation.

When it comes time to transmit data to/from the network, a
dedicated Encode/Decode function is used, to translate the
entire content from its native form to the network form, or
vice versa.

FSW should typically not access the encoded form of data,
outside of the codec routines, except under very limited
circumstances with good reason (such as dynamically updating
the total_length field in the base header after encode).

A significant cleanup and overhaul of CF unit tests
to follow patterns that are more consistent with
other CFE/OSAL test modules.

Since the FSW change in this PR requires significant test updates
to go with it, this replaces all the "cfdp" tests with new
implementations where there is a 1:1 ratio between test
functions and implementation functions.

Ths also does some minor cleanup on the FSW side where needed
for testability.

Note that all stub files in this version are now directly generated
using the tool provided with UtAssert.  These generated stubs should
not be modified.