[SLE15-SP5] Fixed online migration from SLE_HPC to SLES SP6+ (bsc#1220567)

[lslezak]

Problem

• In SP6 the SLE_HPC product is dropped and replaced by SLES + HPC module
• We need to backport the SLE HPC is not a base product anymore (jsc#PED-7841) #650 and [SLE15-SP6] Better summary for the SLE_HPC => SLES upgrade (jsc#PED-7841) #652 pulls to the SLE15-SP5 branch because in online migration (in installed system) the migration to SP6 runs using the original SP5 code. We need to fix the HPC SP5 -> SLES SP6 migration case.
• The problem is that YaST does not recognize the HPC -> SLES migration correctly, it sees installing SLES and removing HPC which is evaluated as a possibly dangerous operation and results in an error in the summary
• https://bugzilla.suse.com/show_bug.cgi?id=1220567

The original error:

Notes

• Additional problem is that the SP5 maintenance update will sooner or later land in the SP5 QU media. And that medium can be used for the SLE_HPC-SP4 -> SLE_HPC-SP5 migration. With the original SP6 patches it would migrate SLE_HPC-SP4 to SLES-SP5 which is not wanted here as we still provide HPC for SP5, the migration to SLES is need only in SP6+.

Solution

• The code need to return a different product mapping for SP5 and SP6 (and newer).
• That also needs a change in the API because that mapping cannot be a constant anymore, it depends on the migration target.
• See also [SLE15-SP5] Set the upgrade product mapping depending on the target version (bsc#1220567) yast-registration#595 which activates the correct product mapping depending on the target version installed
• Additionally I grepped all YaST code to ensure the now private constants are not used elsewhere. They were only used in yast-registration which has been updated in [SLE15-SP5] Set the upgrade product mapping depending on the target version (bsc#1220567) yast-registration#595

Testing

Tested in online (installed system) SP5->SP6 migration in a registered system (using SCC) to test that the original problem is fixed:

Tested in offline (boot from installation medium) SP4->SP5 migration in a registered system (using SCC) to test that there is no regression when the patch gets released in the next QU:

Tested in offline (boot from installation medium) SP4->SP5 migration in a non-registered system (using packages from DVD medium) to test that there is no regression when the patch gets released in the next QU:

[coveralls]

coverage: 37.48% (+0.08%) from 37.405%
when pulling f50efa9 on hpc_sp5
into bc44d3c on SLE-15-SP5.

[yast-bot]

✔️ Internal Jenkins job #755 successfully finished
✔️ Created IBS submit request #323628