clean normalization_inner_outer

[liqiangxl]

similar to #923
(1) Transformed private static functions in the InnerOuterPersistentKernelScheduler class into utility functions within an anonymous namespace.
(2) Removed checks/calculations that are unrelated to inner_outer persistent scheduler.

[liqiangxl]

!build

[liqiangxl]

!build

[liqiangxl]

!build

[liqiangxl]

Checked benchmark cases using tools/compare_codegen.sh -- build/nvfuser_bench --benchmark_filter=NvFuserScheduler --benchmark_repetitions=1 --benchmark_min_time=0

Previous HEAD position was b3051d8f Fix a memory leak. (#936)
Switched to branch 'llu/clean_inner_outer_norm'
Your branch is up to date with 'origin/llu/clean_inner_outer_norm'.


DIFF RESULT:

No difference found

[liqiangxl]

!build

[liqiangxl]

!build

[liqiangxl]

!build

[liqiangxl]

!build

[liqiangxl]

!build

[naoyam]

This part shows as changed. Is there any change here?

Github's diff indicates lots of changes. Maybe it's just false positives? Could you try to minimize the (false?) changes?

[liqiangxl]

No change. I tried hide whitespace and it removes false changes.

[liqiangxl]

!build

[naoyam]

LGTM

[naoyam]

Do you know why the CI diff check shows some diffs?

[liqiangxl]

Do you know why the CI diff check shows some diffs?

The reference used in diff compare is Commits on Sep 24, 2023:
HEAD is now at 4004b72 Store compiled CUDA kernel in Flatbuffers binary (#814)

(1) nvfuser-ci/jit_codegen_diff_10/10 is because more benchmark cases are ran by CI after #930 which is merged on Sep 25.
(2) nvfuser-ci/jit_codegen_diff_5/10 is fixed by #947 which is merged on Sep 26.