* change CUDA implementaion of uniform/gaussian OP

* fix unittest

* force close eager_generator.exe

* modify according to zhouwei's comment

* support vs2019 compilation in windows

* not modify pow_op's original compute logic

* add trt_mobilenet demo test

* only deliver trt_root in run.sh

* modify for wincheck-inference case

* modify according to zhouwei's comment

* open with_trt and block failed unittests in windows

* test

* test on_infer off problem in windows

* turn off on infer in windows-ci

* for pure fp16

* opt topk

* modify for wincheck-inference case

* modify according to zhouwei's comment

* modify according to zhouwei's comment 2nd time
Co-authored-by: Nzhangkaihuo <zhangkaihuo@baidu.com>

move variable UPLOAD_TP_FILE to the beginning or it cant be initialized when running build-whl task (#35895)

* new function: share third party cache among servers to fasten build speed

* modified code according to zhouwei25's comment

* add wget install step, move cd build to the last of if condition

* block note and error of third_party share; change bce upload method

* change third_party sub_dir in bos, since third party in different cuda version cant share

* set sub_dir by get nvcc version

* change third_party local path to be same with bos path

* enable infer_ut on windows

* remove lib calculation & time

* unset http_proxy when download bos file on windows

* polish windows compile for Ninja, fix random compile fail

* polish windows compile for Ninja, fix random compile fail

* Support sccache to speed up compilation on Windows

* Support sccache to speed up compilation on Windows

1. check status of remove op grad shell scripts
2. Update paddle_build.bat to remove op grad when build windows inference lib
3. add inference demo test after build finish

* support Ninja and establish dependencies relationship between paddle with third_party

* fix CI

* support Ninja