* [Plugin] Fix Custom device in eager mode, test=develop

* update test case, test=develop

* update ut for coverage, test=develop

* add xpu resnet_unit
*test=kunlun

* tmp
*test=kunlun

* remove useless code in fc

* reshape fill_constant

* commit

* commit

* transfer as_real to phi

* fix erros

* blocking: True -> False

* add custom-op UTs 1

* add authors
Co-authored-by: NAllen Guo <alleng@graphcore.ai>
Co-authored-by: NZhixin Yao <zhixiny@graphcore.ai>
Co-authored-by: NZhaorui Chen <zhaoruic@graphcore.ai>

* update url
Co-authored-by: NZhixin Yao <zhixiny@graphcore.ai>
Co-authored-by: NZhaorui Chen <zhaoruic@graphcore.ai>

* add custom-op UTs 0

* add authors
Co-authored-by: NAllen Guo <alleng@graphcore.ai>
Co-authored-by: NZhixin Yao <zhixiny@graphcore.ai>
Co-authored-by: NZhaorui Chen <zhaoruic@graphcore.ai>
Co-authored-by: NZhixin Yao <zhixiny@graphcore.ai>
Co-authored-by: NZhaorui Chen <zhaoruic@graphcore.ai>

test=kunlun

* add fused token prune op and plugin

* Fix test_functional_conv2d_transpose random seed

* Fix random seed and use np.testing

* Fix random seed for test_lu_unpack_op

* Fix test_autograd_functional_dynamic random seed

* fix the outputs of net is x,x

* add unittest for duplicate output

* fix

* [CustomDevice] add custom ccl api

* add ut

* hide prim2orig in executor

* add some test cases without param guard

* fix spell error param into program

* Use absolute path when import paddle.incubate.autograd.prim2orig

* Move eigvals OP to PHI

* Fix CI errors

* Fix CI errors

* avoid sync with cpp in partition op

* delay eval & predict mode

* bugfix for gradient merge pass

* add comm init control by socket

* avoid single card instance failure

* add sparse api coalesce