fix: remove parent module use statement in iterative solver submodules

[edvinolo]

Remove the use stdlib_linalg_iterative_solvers statement in the iterative solver implementation submodules. The use statement is not needed since a submodule has access to its parent module via host association. There is also a paragraph in the standard that might disallow this, see e.g. the discussion here: llvm/llvm-project#169863. Flang gives an error for these lines, although apparently not gfortran or ifx.

Along the same lines, the submodules also contain use statements that are already present in the parent module and should therefore be redundant; these could also be removed if you agree.

(@jalvesz @jvdp1 since you have worked on the iterative solvers)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 68.69%. Comparing base (34723c7) to head (45b4825).
⚠️ Report is 11 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1073   +/-   ##
=======================================
  Coverage   68.69%   68.69%           
=======================================
  Files         392      392           
  Lines       12693    12693           
  Branches     1377     1377           
=======================================
  Hits         8720     8720           
  Misses       3973     3973           

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jalvesz]

Thanks for catching this one @edvinolo! I think this can be merged right away.
I'm curious about how far did you manage to go with compiling stdlib with flang?

[jvdp1]

Good catch! thank you @edvinolo

[edvinolo]

@jalvesz With the version of flang-21 on Homebrew for Linux, I was able to compile most of stdlib as is (with only sp and dp enabled). The problems I ran into, other than the one addressed by this PR, were: llvm/llvm-project#169590, and then some of the tests were failing due to this issue: llvm/llvm-project#170331.

I have also tried with a fairly recent development version of flang-22 (also with xdp and qp enabled), but then it crashes when compiling some tests and examples, I think due to an issue with derived type I/O for the string type. I haven't filed an issue in the llvm repo yet. If I skip the tests where flang crashes, then everything compiles, and all the remaining tests pass.

I have to use make as the build backend, if I use ninja with flang, it tries to build some of the files out of order.

[jalvesz]

Brilliant! thanks for the info, looking forward for progress on this