Fix #795, Add uncovered vxworks source files to coverage statistics

[skliper]

Describe the contribution
Fix #795 - Adds all the currently possible source files for vxworks to coverage statistics and adds coverage tests

• Renamed sockaddr* structures to deconflict from structure name in os-impl-bsd-sockets.c to make stubbing out the OS structures easier.
• Compiling os-impl-bsd-sockets.c in coverage test with OS_NETWORK_SUPPORTS_IPV6 to cover that option also
• Renamed bsd-select-stubs.c to sys-select-stubs.c since it's really stubbing sys/select.h not the bsd implementation of select

Testing performed
Build and execute unit tests, passed

Expected behavior changes
None, but coverage now includes all currently possible files in vxworks build

System(s) tested on

• Hardware: cFS Dev Server
• OS: Ubuntu 18.04
• Versions: cFS Bundle main + this commit

Additional context
Not full coverage yet on os-impl-bsd-sockets.c (plan to keep improving prior to CCB)
Also will squash

Third party code
None

Contributor Info - All information REQUIRED for consideration of pull request
Jacob Hageman - NASA/GSFC

[skliper]

@jphickey Is this along the lines of what you'd expect?

[jphickey]

@jphickey Is this along the lines of what you'd expect?

Yes! This looks good!

The only comment is regarding the endian conversions - e.g. ntohl/s, etc - these are ones that I would consider falling back to "real" behavior - similar to memcpy/memset etc - if nothing is configured. Having it just return 0 by default might cause problems for some code. But in those cases I suppose someone can just be sure to set/register a return value in the test case.

[skliper]

...ntohl/s...

I think I left those out since host could be big or little endian, and preference might be a hook that emulates the real target? Just don't want testers to assume coverage testing w/ the test host endian is really appropriate in all cases.

[jphickey]

I do agree that if there is any conditional which depends on output of an endian conversion, that the test case should explicitly set that output, rather than assuming the stub is going to work like the real one does. So it's probably fine the way it is.

[astrogeco]

CCB:2021-02-24 APPROVED

• Doesn't get to 100% coverage yet

[skliper]

See #831 for follow-on to finish adding coverage

[skliper]

@astrogeco squashed, ready for merge