gfortran: declare mapper is not accepted ("Unclassifiable OpenMP directive")

! # Finding (GCC 16.2): gfortran does not accept declare mapper; the reproducer is the catalog program omp52-clause-map-mapper
!
! - toolchains: gcc_g90 (GCC 16.2.0, AMD Instinct MI250X gfx90a, Tioga) and gcc_sm80 (GCC 16.2.0,
!   NVIDIA A100 sm_80, wingtip); separate memory only; -O2; OMP_TARGET_OFFLOAD=mandatory
! - build: gfortran -fopenmp -foffload=amdgcn-amdhsa -foffload-options=amdgcn-amdhsa=-march=gfx90a -O2 (gcc_g90);
!   -foffload=nvptx-none -foffload-options=nvptx-none=-march=sm_80 -foffload-options=nvptx-none=-latomic -O2 (gcc_sm80)
! - failure: rejected: Unclassifiable OpenMP directive
! - observed: `Error: Unclassifiable OpenMP directive at (1)` at the declare mapper directive, on both GPUs; every Fortran map, to
!   and from program with a mapper fails the same way.
! - expected: compiles; `RESULT ran_on_host=0 check=pass`
! - GCC 16.2 documentation (libgomp.texi of the GCC 16.2.0 release tarball, "OpenMP Implementation Status", OpenMP 5.0 table):
!   "declare mapper directive": P, "Initial support and for C/C++, only".
! - spec: the declare mapper directive (5.2 §5.8.8 p.162).
! - other families: AFAR runs the program correctly; CCE 21 runs it (CCE 19/20 lose a copy-back after the mapper,
!   cce-mapper-next-map-copy-back.f90); NVHPC rejects it.
! - gate and confirmer: 18 cells (9 gcc_g90, 9 gcc_sm80), each confirmed in the gate's four environments (a-d;
!   results/gcc/gate-tioga/ and gate-wingtip/ verdicts.jsonl); confirmer verdicts (results/gcc/confirm/batch.gcc_*.jsonl):
!   compiler_bug (R11-compile-rejection) 12, unsupported (R11a-newer-form-unclaimed) 6
! - catalog programs (9): omp50-clause-from-mapper, omp50-clause-map-mapper, omp50-clause-to-mapper, omp52-clause-from-mapper,
!   omp52-clause-map-mapper, omp52-clause-to-mapper, omp60-clause-from-mapper, omp60-clause-map-mapper, omp60-clause-to-mapper
! - legality: the catalog programs are legal by construction (offload.engine.gen, c_gen and acc_gen
!   check every program against the catalog rules and the offload scope as it is built; no
!   compiler). GCC is a compiler under test only: never a legality witness, here or for another
!   family's case (catalog/machines.yaml under_test_only).
! Generated by compliance.gen (OpenMP 5.2, free-form Fortran). Do not edit.
! id: omp52-clause-map-mapper
! primary atom: omp.clause.map.mapper
! legal: yes
! observable: yes
! ran_on_host: inside the target region of interest
! evidence: copy_mode_only
! needs: none
! std: f2008
! derivation (check=pass): mapper mm maps q%v with map type to; map(mapper(mm), tofrom: s) applies it with the decayed map type:
! mapper to with clause tofrom gives to (5.2 §5.8.8 p.163 (Table 5.3)); so s%v (1) is copied to the device and not back: the device
! writes b(i) = 2*a(i) + s%v = 2*i + 1 = ref(i), then s%v = 7 on the device only; in copy mode the host's s%v stays 1 (the
! copy-mode-only check), under shared storage that check is skipped; the check accepts
program gen_prog
  use omp_lib, only: omp_is_initial_device
  use, intrinsic :: iso_c_binding, only: c_loc
  implicit none
  type :: dt
    integer :: v
  end type dt
  !$omp declare mapper(mm: dt :: q) map(to: q%v)
  integer, parameter :: n = 64
  integer, allocatable, target :: a(:), b(:)
  integer :: ref(n)
  integer :: i, host_flag
  logical :: ok
  type(dt) :: s
  logical :: pres, shared_mem
  integer :: zc(1)
  allocate(a(n), b(n))
  do i = 1, n
    a(i) = i
    b(i) = 0
    ref(i) = 2*i + 1
  end do
  host_flag = 1
  s%v = 1
  pres = .true.
  zc = 0
  !$omp target map(to: zc)
    zc(1) = 1
  !$omp end target
  shared_mem = zc(1) == 1
  !$omp target map(to: a) map(from: b) map(mapper(mm), tofrom: s) map(tofrom: host_flag)
    host_flag = merge(1, 0, omp_is_initial_device())
    do i = 1, n
      b(i) = 2*a(i) + s%v
    end do
    s%v = 7
  !$omp end target
  if (.not. shared_mem) pres = pres .and. (s%v == 1)
  ok = all(b == ref) .and. pres
  print '(a,i0,a,a)', 'RESULT ran_on_host=', host_flag, ' check=', merge('pass', 'fail', ok)
  deallocate(a, b)
end program gen_prog