build: consume Fortio lifecycle fix

[krystophny]

Summary

Update all libneo Fortio consumers to immutable Fortio lifecycle-fix commit 11afd0b, including CMake, fpm, and h5merge.

This keeps libneo standalone through Fortio; it does not add a system HDF5 link. The Fortio close semantics now match the normal HDF5 contract and keep the KAMEL append gate at native speed.

Upstream Fortio PR: lazy-fortran/fortio#68

The ITP-plasma consumer pins in the matched PRs are advanced to this bridge commit.

[krystophny]

This FortFEM-specific bridge remains for MEPHIT #37. Main-line consumers such as SIMPLE, NEO-2, NEO-RT, rabe, and MEPHIT #41 now use the compatible main-based bridge in PR #414: #414. Both bridges consume standalone Fortio PR #68.

[GeorgGrassler]

Can not say much to the specifics, the cuts/updates look sane. A comment just regarding the test associated with netcdf and hdf5 in libneo: Should nf-config still be used (

libneo/test/CMakeLists.txt

Line 42 in f05e4f2

execute_process(COMMAND nf-config --flibs

)? Cause as far as I get, you have removed the netcdff lib now from the ci, so probably this should also go? If yes, I recommend to sweep the test suite once more for other dead parts as everything is now fortio.

Also, are the removed symlinks an unrelated house-keeping? Fine if, just wanna know for completeness.

[krystophny]

Closing as obsolete rather than merging this conflicting rollout branch. The I/O migration already landed in #407 and the lifecycle pin landed on current main in #414. MEPHIT #37 has now been updated to the merged #414 commit (f05e4f20ce32dbb0b9e4b5e3c433fe6a2b8ce770); its FortFEM-enabled build succeeds and all five registered tests pass.

@GeorgGrassler On the review questions: the removed one-line source aliases were intentional migration work to avoid duplicate fpm sources, not unrelated housekeeping. The remaining nf-config --flibs call in test/CMakeLists.txt appears to be dead test-build residue and should be removed separately from current main with focused verification.