[ hgemm ] Use hgemm kernel in transpose cases

[skykongkong8]

• With SIMD version of fp16 transpose code, using hgemm kernel in transpose case would be more useful.
• Note that we should develop a data packing code for this case for further optimization.
• Still, this gives huge difference btw

hgemm_transA

dim	before	after
1024 square	192 ms	35 ms
768 square	71 ms	13.0 ms
256,1440,256	15 ms	2.8 ms
256,256,1440	15 ms	3.0 ms
1920,1560,88	46 ms	9.8 ms

hgemm_transB

dim	before	after
1024 square	88 ~ 113 ms	30 ms
768 square	31 ms	12 ms
256,1440,256	6.4 ms	2.6 ms
256,256,1440	7.1 ms	3.0 ms
1560,2048,512	103 ~ 188 ms	45 ms

hgemm_transAB

dim	before	after
1024 square	2599 ms	37 ms
768 square	814 ms	12 ms
256,1440,256	160 ms	2.8 ms
256,256,1440	70 ms	3.0 ms

Self evaluation:

1. Build test: [X]Passed [ ]Failed [ ]Skipped
2. Run test: [X]Passed [ ]Failed [ ]Skipped

[taos-ci]

:memo: TAOS-CI Version: 1.5.20200925. Thank you for submitting PR #2645. Please a submit 1commit/1PR (one commit per one PR) policy to get comments quickly from reviewers. Your PR must pass all verificiation processes of cibot before starting a review process from reviewers. If you are new member to join this project, please read manuals in documentation folder and wiki page. In order to monitor a progress status of your PR in more detail, visit http://ci.nnstreamer.ai/.

[taos-ci]

:octocat: cibot: @skykongkong8, A builder checker could not be completed because one of the checkers is not completed. In order to find out a reason, please go to http://ci.nnstreamer.ai/nntrainer/ci/repo-workers/pr-checker/2645-202406181327260.34216904640198-351ee4214fa04fa85a4ad56871208892061ff015/.

[taos-ci]

:octocat: cibot: @skykongkong8, A builder checker could not be completed because one of the checkers is not completed. In order to find out a reason, please go to http://ci.nnstreamer.ai/nntrainer/ci/repo-workers/pr-checker/2645-202406191313300.53744697570801-6371277f9ff5fda46b77dc762ad000921e211828/.