[refactor][arm server] move infer_dims_func_ & infer_type_func_ regis… …ter code to constructor
[build] update deps and release v0.7.0
[build] export version info for integration
[feature][cuda] add DataVisitor to access constants
[cuda] Add register/remove algo and fusion
[fix] demo compilation
[cuda][fix] change num_chl_per_grp to num_flt_per_grp and num_flt to num_flt_pad
[fix][x86] fix graph optimize rule `FuseConvEltwise`
[doc] update docs about compiling CUDA engine on Windows
[python] compilation fix