Merge to master the fix about wrongly allocated RAID partitions

[ancorgs]

This is a merge of #2854 into the master branch.

Bonus: This also includes a change in one of the existing tests to make it more robust. See the comment below.

[joseivanlopez]

LGTM

[joseivanlopez]

Some tests are failing.

[ancorgs]

The failed tests are not related to this change, but to recent changes in yast2-storage-ng. But I will take the opportunity to fix that as well.

[coveralls]

coverage: 64.151% (-11.7%) from 75.85%
when pulling 843ed00 on bug_partitioned_raid_master
into 52366f5 on master.

[ancorgs]

Regarding the failed test.

The root of all evil is that we do not mock the value of StorageEnv.instance.no_bls_bootloader, which is forced to true during the real Agama execution but not set (mocked) on this test.

But even knowing that in advance, I was curious to check why the result is sometimes green (for example in the SLE-16 branch CI or in my laptop) and sometimes red (like in the master branch CI and in my workstation). I found the cause now.

Since yast/yast-storage-ng#1413, the content of /etc/YaST2/control.xml can influence the result of the test. If there is an entry for preferred_bootloader with the value "grub2-bls" (something that happens in recent instances of TW or Slowroll) then the test will be red. If there is no such an entry (something that happens in any other distribution or in not-so-recent TW instances) the test will be green.

Having solved the mystery, now I will fix the real cause of the instability of this test - the lack of mocking for StorageEnv.instance.no_bls_bootloader.