cpu: x64: improve memory management safety for additional kernels

[Permanence-AI-Coder]

Description

The code changes in this PR were generated automatically by the Permanence AI Coder and reviewed by @fwph & @jweese. This PR updates memory allocation for jit_generator objects across several structs. Note that in the case of jit_avx512_core_x8s8s32x_fwd_kernel the macro DNNL_DISALLOW_COPY_AND_ASSIGN was missing; and was added. This PR is effectively applying the update from https://github.com/oneapi-src/oneDNN/pull/2017 to additional files.

on-behalf-of: @permanence-ai [email protected]

Checklist

General

• [X] Do all unit and benchdnn tests (make test and make test_benchdnn_*) pass locally for each commit?
• [X] Have you formatted the code using clang-format?

[fwph]

Hi @mgouicem, just a reminder to check the restored macros!