*Pool2d supports adaptive param

* improve 3x3s1 direct profile

* fix format, test=develop

fix clang build error

improve sgemm performance on A53

1. fix clang android log warning.
2. remove an unused event.

* [lite][opencl] remove event with clfinish, add strict check for cl warning. add conv 3x3opt fallback opt layout cast ,test=develop

* [LITE][OPENCL]rm event in element_add_buffer_compute test=develop

* [LITE][OPENCL]suite cl_functions_test.cc test=develop

* [LITE][OPENCL] suite cl_common.sh lint check test=develop

* [LITE][OPENCL] suite conv_image_compute.cc lint check test=develop

* [LITE][OPENCL] suite cl_wait_list() lint check test=develop

* fix bilinear opencl kernel. test=develop

* [LITE][OPENCL] replace map with memsync. test=develop

refactor(*): reduce Wsign-compare warning

for multiple-use args, only cast once

* Optimize matmul for size(x_dims)=2  size(y_dims)>2

* [NPU] add shape bridge

move shape arm kernel to host

* enhance compare arm kernel

* [NPU] add gather op bridge

* enable reshape arm ut

* [NPU] add lookup_table bridge

lite cuda support exec multi-stream

* [LITE][OPENCL][Image] lazy init cl::event , test=develop

* [LITE][OPENCL][Image] lazy init cl::event , test=develop

[LITE][OPENCL][Image] use unique_ptr to wrap filter and bias in conv image compute , test=develop (#3388)

* [LITE][BM] fix reshape infer shape issue, test=develop

* [LITE][BM] with testing=on, test=develop

1. clean code;
2. change `cl::Kernel` from unique to shared ptr;
3. `reset` `cl::Program` and `clear` `device_info_` in destroyed of CLRuntime;
4. remove clFlush in destroyed of CLRuntime.

fix opencl hang on mali

* Optimize weight quantizaion, test=develop