split_single_dyn_dim: add bucket_by_optimals to cut dyn-shape compile…

[chun-wan]

• split_single_dyn_dim gains a bucket_by_optimals pass field. When true and
  a dynamic dimension has non-empty optimals, it emits one static submodule
  per (min, optimals..., max) instead of one per integer in [min, max].
• At runtime select_module::compute() first tries an exact-shape match, then
  falls back to the smallest compatible submodule (same type/rank, every bucket
  dim >= the input dim). On the ref backend the input is zero-padded into the
  bucket-sized buffer on the host; GPU callers pre-pad.
• GPU/Ref targets read MIGRAPHX_DYN_DIM_BUCKET_BY_OPTIMALS once and forward it
  to the pass field. The field is a normal pass parameter, so unit tests set it
  directly via run_pass(...) rather than env vars.
• Tests: test/split_single_dyn_dim_test.cpp (bucketed enumeration) and
  test/ref/select_module.cpp (exact match / round-up-with-host-pad /
  too-big-throws).
• Split out of Reduce dynamic-shape compile cost and select_module dispatch overhead #4911; the per-instance select_module dispatch cache (a pure
  runtime optimization) is a separate follow-up PR.

[github-actions]

Thank you for your contribution! Since this is an external pull request, a maintainer must review PR and add the "ok-to-test" label if it is approved for testing.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #5028      +/-   ##
===========================================
+ Coverage    92.89%   92.89%   +0.01%     
===========================================
  Files          603      603              
  Lines        32448    32482      +34     
===========================================
+ Hits         30140    30174      +34     
  Misses        2308     2308              

Files with missing lines	Coverage Δ
src/include/migraphx/op/select_module.hpp	90.91% <100.00%> (+4.37%)	⬆️
src/include/migraphx/split_single_dyn_dim.hpp	100.00% <ø> (ø)
src/split_single_dyn_dim.cpp	98.65% <100.00%> (+0.19%)	⬆️
src/targets/ref/target.cpp	100.00% <ø> (ø)

🚀 New features to boost your workflow:

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

[chun-wan]

This is the lower-risk half of #4911, split out per review feedback so it can land independently. It contains only the compile-time bucketing: split_single_dyn_dim{.bucket_by_optimals} (a pass parameter, driven directly in tests) plus select_module's smallest-compatible-bucket dispatch with host padding. The per-instance dispatch cache from #4911 is intentionally not here — it'll be a separate follow-up PR. All CI is green except security_gate (fork-only). @pfultz2 @CharlieL7 could you review? Thanks!

[chun-wan]

This is the low-risk, bucket-only split out of #4911: just MIGRAPHX_DYN_DIM_BUCKET_BY_OPTIMALS in split_single_dyn_dim + a smallest-compatible-bucket fallback in select_module (no freeze pass, no dispatch cache). Just rebased onto latest develop; all code checks pass, only security_gate (infra) is red.

It collapses dynamic-shape compile from O(max−min) → O(|optimals|) — measured on a customer model: max=1024 → 4.3s, max=100000 → 5.1s, vs the legacy path not finishing in 150s.

@CharlieL7 @kahmed10 could one of you review when you have a chance? 🙏