Add quantized gelu (#119935)
Summary: Added Quantized gelu for vulkan backend.
Test Plan:
**Tested it on "On Demand RL FBSource"**
LD_LIBRARY_PATH=third-party/swiftshader/lib/linux-x64/ buck2 run fbcode/mode/dev-nosan //xplat/caffe2:pt_vulkan_quantized_api_test_bin -c pt.vulkan_full_precision=1 -- --gtest_filter="VulkanAPITest.gelu_q*"
----------------------------------------------------------------------------------
Note: Google Test filter = VulkanAPITest.gelu_q*
[==========] Running 4 tests from 1 test suite.
[----------] Global test environment set-up.
[----------] 4 tests from VulkanAPITest
[ RUN ] VulkanAPITest.gelu_qint8
[ OK ] VulkanAPITest.gelu_qint8 (318 ms)
[ RUN ] VulkanAPITest.gelu_qint8_self
[ OK ] VulkanAPITest.gelu_qint8_self (214 ms)
[ RUN ] VulkanAPITest.gelu_quint8
[ OK ] VulkanAPITest.gelu_quint8 (152 ms)
[ RUN ] VulkanAPITest.gelu_quint8_self
[ OK ] VulkanAPITest.gelu_quint8_self (142 ms)
[----------] 4 tests from VulkanAPITest (828 ms total)
[----------] Global test environment tear-down
[==========] 4 tests from 1 test suite ran. (828 ms total)
[ PASSED ] 4 tests.
Differential Revision: D52985437
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119935
Approved by: https://github.com/jorgep31415
