Test metadata templating

[comps]

Currently, we copy/paste per-profile definitions like this:

main.fmf snippet

/anssi_bp28_high:

/anssi_bp28_enhanced:
    tag+:
      - subset-profile

/anssi_bp28_intermediary:
    tag+:
      - subset-profile

/anssi_bp28_minimal:
    tag+:
      - subset-profile

/bsi:
    adjust+:
      - when: distro != rhel-9
        enabled: false
        because: BSI profile is currently only on RHEL-9

/cis:

/cis_server_l1:
    tag+:
      - subset-profile

/cis_workstation_l2:

/cis_workstation_l1:
    tag+:
      - subset-profile

/cui:
    tag+:
      - fips
    adjust+:
      - when: distro >= rhel-10
        enabled: false
        because: there is no CUI profile on RHEL-10+

/e8:

/hipaa:

/ism_o:
    tag+:
      - fips

/ospp:
    tag+:
      - fips
    adjust+:
      - when: distro >= rhel-10
        enabled: false
        because: there is no OSPP profile on RHEL-10+

/pci-dss:

/stig:
    tag+:
      - fips

/stig_gui:
    adjust+:
      - enabled: false
        because: not supported without GUI, use stig instead

/ccn_advanced:
    adjust+:
      - when: distro == rhel-8 or distro == rhel-10
        enabled: false
        because: CCN profiles are not present on RHEL-8 and on RHEL-10

/ccn_intermediate:
    tag+:
      - subset-profile
    adjust+:
      - when: distro == rhel-8 or distro == rhel-10
        enabled: false
        because: CCN profiles are not present on RHEL-8 and on RHEL-10

/ccn_basic:
    tag+:
      - subset-profile
    adjust+:
      - when: distro == rhel-8 or distro == rhel-10
        enabled: false
        because: CCN profiles are not present on RHEL-8 and on RHEL-10

and that is not sustainable when CaC/content upstream keeps adding more profiles.

At the same time, FMF doesn't natively support any kind of multi-file templating, and while TMT has something in the works (they call it "policy", I believe, teemtee/tmt#3538), it doesn't work on generic FMF metadata.

As such, maybe we should create our own templating logic, perhaps using python code in scripts/*.py as configured by some conf/*.py file, that would generate / alter main.fmf files across the hierachy, adding per-profile tests.

Maybe the solution would be something like

1. conf/metadata.py having
   1. a python list of .fmf file paths to modify
   2. a python list/dict/whatever of metadata to add, while NOT using any lib.* functionality, to make the metadata-building independent of OS, so the generated code is still fmf/tmt compatible
2. Github CI deployment logic that automatically builds and deploys pre-built Contest somewhere, so we would have deterministic location to fetch built Contest from in CI pipelines
3. Some way to build this manually, maybe a Makefile or manual ./scripts/build_metadata.py invocation, or whatever

Maybe we could store the configuration in the building script itself, instead of conf/, if conf/ seems too weird for something that external to the test suite.

[comps]

Instead of modifying main.fmf, we could maybe generate per-profile *.fmf files in the same directory as the base main.fmf - from FMF PoV, the effect is the same, except we don't need to magically modify a pre-existing file.
The new files could be easily .gitignore'd.

They could also be just symlinks to some shared per-profile definitions, so the script would basically just re-link all files from some shared dir into all of the per-test dirs.

We might need to turn some tests into directories (ie. with-gui from with-gui.fmf into with-gui/main.fmf, etc.) to avoid duplicate file names, but that's fine.
TODO: If we do this, add a Github CI test to ensure there are no non-main.fmf ever added by people.