Pad: Add support for all datatypes in opset-11 spec (#4021)
* Pad: Add support for all datatypes in opset-11 spec
Pad opset-11 implementation supports:
int32, int64, float & double
Per specification, Pad opset-11 also supports:
uint8, uint16, uint32, uint64, int8, int16 & float16
This commit add support for those types to get full coverage of Pad opset-11 operator.
* Pad: Remove 16-bit datatypes support
These types are unused at the moment and binary size is impacted. Remove support for those type to lower binary size.
