[GPU] set mem_flags::need_blocked for conv/deconv postops

[sungeunk]

Description of the issue(symptom, root-cause, how it was resolved)

convolution in/out tensor
input tensor: f16:b_fs_yx_fsv16:1x512x78:nopad
input tensor: f16:bfyx:512x512x3:nopad
input tensor: f16:bfyx:1x512x1:nopad
input tensor: f16:b_fs_yx_fsv16:1x512x78:nopad --> binary_add
input tensor: f16:b_fs_yx_fsv16:1x1x1:nopad --> binary_mul
output tensor: f16:b_fs_yx_fsv16:1x512x78:nopad

• onednn memory desc of conv/deconv needs setting onednn::mem_flags::need_blocked if the format is blocked.

The code and line that caused this issue (if it is not changed directly)

• src/plugins/intel_gpu/thirdparty/onednn_gpu/src/common/primitive_attr.cpp (validate_binary)

Reproduction step and snapshot (model is attached in the ticket)

• benchmark_app.exe -d GPU --hint latency -niter 1 -m kokoro.xml -shape "ref_s[1,256],input_ids[1,78]"

Problematic graph

• This issue could not check with graph.

Checklist

• Is it a proper fix? (not a workaround)
• Did you include test case for this fix, if necessary?
• Did you review existing test that can be extended to cover this scenario? Which test did you review?
  -- unit/func test can not cover this case due to target shape is 3d and blocked format.

Tickets:

• 169876

[jade-cho]

LGTM