This test has been occasionally failing under spread:
```
FAIL: fde_test.go:492: fdeSuite.TestRevealErr
using shellcheck: ""
fde_test.go:508:
c.Assert(err, ErrorMatches, `(?s)cannot run fde-reveal-key "reveal":
-----
failed
service result: exit-code
-----`)
... error string = "cannot run fde-reveal-key \"reveal\": exit status 1"
... regex string = "" +
... "(?s)cannot run fde-reveal-key \"reveal\": \n" +
... "-----\n" +
... "failed\n" +
... "service result: exit-code\n" +
... "-----"
OOPS: 9 passed, 5 skipped, 1 FAILED
```
While we couldn't yet reproduce it locally and find what exactly
triggers it, it makes sense to exclude systemd-run from the picture by
mocking it. We now simply check that it was invoked with the right
parameters and that its output is captured correctly.
Adjusted code to the refactoring using kernel/fde.
Split responsibility to deal with v1/v2 hooks with kernel/fde.
Now dealing with aux key will come later.
this stops kernel/fde from importing secboot, which also means that
secboot can now kernel/fde naturally
rename/reorg/split secboot_tpm.go into hooks specific bits
and general secboot-using secboot_sb.go
This commit changes the supported yaml for the kernel snap. For
the raspberry pi DTB use case the kernel assets are tightly
coupled with the kernel. So the edition would have to be bumped
everytime the kernel is build. So the edition does not make much
sense in this context. Hence a new "update" field that is boolean
for now but we may expand it later into a map. This map would
allow to specify what content items should get updated and which
should not get updated.
