POC: allow multi-level/subgraph duplication of kernels

[MichaelSt98]

POC duplicating a kernel and optionally duplicating/separating the underlying subgraph.

E.g., starting with:

the result of the DuplicateKernel trafo with duplicate_kernels=('kernel',) is:

However, with this PR one can specify duplicate_subgraph=True, which yields for the example above:

Further, the actual transformation part works as well, producing e.g.,:

MODULE kernel_mod_duplicated
  IMPLICIT NONE
  CONTAINS
  SUBROUTINE kernel_duplicated (klon, field1)
    USE kernel_nested_mod_duplicated, ONLY: kernel_nested_vector_duplicated, kernel_nested_seq_duplicated
    USE compute_2_mod_duplicated, ONLY: compute_2_duplicated
    IMPLICIT NONE
    INTEGER, INTENT(IN) :: klon
    INTEGER, INTENT(INOUT) :: field1(klon)
    INTEGER :: tmp1(klon)
    INTEGER :: jl
    
    CALL kernel_nested_vector_duplicated(klon, field1)
    
    DO jl=1,klon
      CALL kernel_nested_seq_duplicated(field1(jl))
      CALL compute_2_duplicated(field1(jl))
      tmp1(jl) = 0
      field1(jl) = tmp1(jl)
    END DO
    
  END SUBROUTINE kernel_duplicated
END MODULE kernel_mod_duplicated

[github-actions]

Documentation for this branch can be viewed at https://sites.ecmwf.int/docs/loki/474/index.html

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.05%. Comparing base (c054b46) to head (7dff96e).
Report is 4 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #474      +/-   ##
==========================================
+ Coverage   96.03%   96.05%   +0.01%     
==========================================
  Files         226      226              
  Lines       40579    40639      +60     
==========================================
+ Hits        38970    39035      +65     
+ Misses       1609     1604       -5     

Flag	Coverage Δ
lint_rules	96.39% <ø> (ø)
loki	96.04% <100.00%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[reuterbal]

We need to consider whether we want to directly provide the SGraph to transformations or build an API to semantically capture the changes and have the scheduler apply them afterwards.
The latter seems more elegant (not hacking with core data structures) but it might be tricky to build something equally flexible.
Also note that tthe graph may be a temporary object here when traversing a file graph. In that case, this would not work at all.