[v1.5.x] Backport to [MKL-DNN] Integrate Conv3d and Pool3d/1d (#17884)

[ChaiBapchya]

• Integrate MKl-DNN conv3d and pool3d/1d

• fix UT & address comments

• clean code

• rebase against latest master

[mxnet-bot]

Hey @ChaiBapchya , Thanks for submitting the PR
All tests are already queued to run once. If tests fail, you can trigger one or more tests again with the following commands:

• To trigger all jobs: @mxnet-bot run ci [all]
• To trigger specific jobs: @mxnet-bot run ci [job1, job2]

---

CI supported jobs: [unix-gpu, centos-gpu, website, edge, clang, windows-gpu, centos-cpu, miscellaneous, unix-cpu, windows-cpu, sanity]

---

Note:
Only following 3 categories can trigger CI :PR Author, MXNet Committer, Jenkins Admin.
All CI tests must pass before the PR can be merged.

[TaoLv]

Hi @ChaiBapchya, probably you cannot do this kind of back porting. I see the v1.5.x branch still depends on MKL-DNN v0.19. But the new Conv3D and Pool3D integration is developed based the master branch of MXNet and v1.x of DNNL (now oneDNN). There are many incompatible changes from v0.19 to v1.x of the library. See #16555 for more details.

[TaoLv]

This feature will be released in the coming v1.7.0 release.

[ChaiBapchya]

@TaoLv Thanks for the clarification. Closing for now.

[ChaiBapchya]

@TaoLv since the customer ask is for v1.6.0 is it possible to backport this fix to that branch? 1.6 seems to have updated mkldnn version.

[ChaiBapchya]

Trying #18206