fwts fails if OPBUILD_VENDOR is set, documentation is inconsistent

[AlexanderAmelkin]

We're experiencing a problem with vesnin that fwts test suite fails at this call.

The reason for this is that we redefine OPBUILD_VENDOR to vesnin in accordance to VERSION.readme.

That results in open-power string to disappear and get replaced with yadro and op-build strings, with the latter being incorrectly parsed by hostboot for P8 resulting in an op=build-... attribute of firmware node of the device tree. That's something that fwts apparently doesn't expect and that is understandable, because that's not what one would expect after reading the VERSION.readme file.

It appears like the intention was to preserve the op-build version if OPBUILD_VERSION is set, but in fact the op-build string is created when OPBUILD_VENDOR is defined.

We would propose to:

• Always keep the open-power string, regardless of OPBUILD_VENDOR;
• When OPBUILD_VERSION is not defined, use the default value in open-power string;
• When OPBUILD_VERSION is defined, create an opbuild string (without hyphen) with the default value for open-power; Alternatively, use op-build with hyphen, but add support for that into hostboot/skiboot/whatever other part parses the strings;
• Get rid of OPBUILD_VENDOR variable, don't create a string for it: the vendor name is available in the top node of device tree if the main board VPD is properly filled, the vendor name string inside the list of versions looks weird, and there is no other use for OPBUILD_VENDOR variable;