Merge latest develop into meta/fmha-fwd-block-sparsity (resolve conflicts)

[poyenc]

Summary

Resolves the merge conflicts blocking PR ROCm#6534 by merging the latest ROCm/rocm-libraries:develop into this branch.

The branch had diverged: 39 commits ahead of a merge base that develop was ~5455 commits past. Three files conflicted, all in the FMHA forward example/test:

• example/ck_tile/01_fmha/example_fmha_fwd.cpp
• example/ck_tile/01_fmha/fmha_fwd_runner.hpp
• test/ck_tile/fmha/test_fmha_fwd.cpp

Resolution

All conflicts were the same "both-added" shape: this branch added a block_mask parameter while develop independently added pack_gqa. Both were kept, with two correctness details:

• In the .insert(...) fluent arg-parser chain, dropped the statement-terminating ; after block_mask so the chain continues into pack_gqa.
• Ordered arguments block_mask_str -> pack_gqa at every call site to match the runner signature in fmha_fwd_runner.hpp (positional args).

Formatting

• C++ files formatted with clang-format-18 using projects/composablekernel/.clang-format.
• No Python files were touched.

Verification

• Zero conflict markers remain.
• git diff of the merge result vs origin/develop shows only this branch's block_mask additions; develop's pack_gqa preserved.

How to land

Merge this into meta/fmha-fwd-block-sparsity; that advances the head branch of ROCm#6534 and clears its conflict state against develop.