arm unified elempack optimization for groupnorm

[mmyyy22]

• add groupnorm_arm optimize

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 99.29245% with 3 lines in your changes missing coverage. Please review.

Project coverage is 95.70%. Comparing base (1d84e98) to head (e7e770e).
Report is 2 commits behind head on master.

Files with missing lines	Patch %	Lines
src/layer/arm/groupnorm_arm.cpp	99.20%	2 Missing ⚠️
src/layer/arm/groupnorm_arm_asimdhp.cpp	99.41%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##           master    #4080    +/-   ##
========================================
  Coverage   95.70%   95.70%            
========================================
  Files         827      829     +2     
  Lines      269904   270328   +424     
========================================
+ Hits       258301   258722   +421     
- Misses      11603    11606     +3     

☔ 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.

[nihui]

hi, the test failure indicates that your code produces wrong result in some condition. please investigate it.

[tencent-adm]

Thank you for your submission, we really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 2 committers have signed the CLA.

✅ nihui
❌ mmyyy22
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[Copilot]

Pull Request Overview

This PR introduces an optimized group normalization implementation for ARM with unified elempack support, enhancing performance in FP16 calculations.

• Updated test logging to include additional dimensions and the affine flag.
• Added a new implementation (groupnorm_arm_asimdhp.cpp) using ARM NEON intrinsics for FP16 arithmetic.
• Introduced header declarations in groupnorm_arm.h for the new methods.

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File	Description
tests/test_groupnorm.cpp	Updated test log formatting to print additional tensor dimensions and the affine flag.
src/layer/arm/groupnorm_arm_asimdhp.cpp	New optimized FP16 group normalization implementation using ARM NEON intrinsics.
src/layer/arm/groupnorm_arm.h	Header declarations for the ARM-optimized groupnorm functions.

Comments suppressed due to low confidence (1)

src/layer/arm/groupnorm_arm_asimdhp.cpp:25

• Consider adding a function-level comment for groupnorm_fp16s to explain its purpose, algorithm, and parameter assumptions. This will improve code readability and maintainability.

static void groupnorm_fp16s(__fp16* ptr, const float* gamma_ptr, const float* beta_ptr, float eps, int channels, int size, int elempack, size_t cstep)

[nihui]

Thanks for your contribution !

[github-actions]

The binary size change of libncnn.so (bytes)

architecture	base size	pr size	difference
x86_64	16495480	16495480	0 😘
armhf	7369732	7380016	+10284 ⚠️
aarch64	10775832	10777936	+2104 ⚠️