* simplify binaryop

* less gpu test

* update binaryop broadcast doc

* do not test atan2 zero

* fix overwrite builtin layer destruction

* make modelbin class copyable

* test++

* force global cpu info initialization

* sanitize zero nT

* fix vulkan validation error, prefer VK_KHR_buffer_device_address over VK_EXT_buffer_device_address

* enable validation extension features

* workaround moltenvk error on spec const composite op

* workaround moltenvk crying on binding image with memory offset

* fix aarch64 build without fp16 conversion intrinsics

* vfpv4 always implies neon

* move x86 abs_mathfun to x86_mathfun

* Improve the implementation of AVX and AVX512F abs.

* Unified naming rules

---------

Co-authored-by: MouriNaruto <Mouri_Naruto@Outlook.com>

* Add some extension capability for vma

* early return for cpu test

* make nvidia driver happy

* fix gemm x86 threading

* new api for detecting arm midr and a53 a55 arch info wrapper

* let a35 be a53 :P

* a53 bf16s

* detect running core

* arm deconv matmul use gemm

* reduce gemm armv7 register uses

* mha arm use softmax fp16

* fuse im2col and packb tile