Previously, pdftk (part of the ticket, badge, ... generation pipeline)
would fail with:
```
Error occurred during initialization of VM
Failed to mark memory page as executable - check if grsecurity/PaX is enabled
```
Thise caused pdf generation to fail.
Since pdftk is a java application and, according to systemd.exec(5),
> Note that [MemoryDenyWriteExecute=] is incompatible with programs and
> libraries that generate program code dynamically at runtime, including
> JIT execution engines, executable stacks, and code "trampoline" featu
> re of various C compilers.
Disabling `MemoryDenyWriteExecute=` fixes it.
As a follow-up to https://github.com/NixOS/nixpkgs/pull/294096 this
should further improve the flexibility around building OS images with
systemd-repart:
* Previously the attribute set `compression` needed to be fully
populated, including `algorithm` and `level` because
`compression.enable` was evaluated by bash, after being interpolated
as strings into the `buildCommand`. Now it's sufficient to pass
`compression.enable = false` to the builder, e.g. in `overrideAttrs`,
to disable the compression.
* Using mkDerivation allows for much more customization than the
previously used `runCommand`, making use of phases and pre/post hooks.
This is especially helpful for building multiple images from the same
system configuration, e.g. to build an image `Y` based on a partially
built raw image `X`, by injecting a UKI that depends on `X` into a
defered ESP.
* Before this change it was non-trivial to conduct further manipulations
on the amended repart definitions. Now, the definitions that
systemd-repart uses to build the image can be easily manipulated in
`postPatch` or `preBuild`.
Aside from this, the build is now executed in the build directory, rather
than `$out`. This allows references to relative paths in the build
environment to be used, especially for `--definitions`, which previously
required an absolute path.
This option makes it easier to reuse a system's ukify.conf without the
need for manually calling the generator on `settings` again to receive a
rendered configuration file.
Theoretically, a complete configuration file could now be provided by
users.
The original b-c-c subtest is probably flaky, as I didn't actually ensure
the inputbox focused before typing. Instead of another hacky xdotool,
split the run dialog subtest out.
https://hydra.nixos.org/build/253473303/nixlog/37
