* Support Mask-RCNN from TensorFlow

* Fix a sample

The while loop would run only once making it useless and leading
to confusion.

So, remove the unwanted while loop and just keep an infinite for
loop.

Intrinsics must be effective, so don't declare FP16 type/operations if there is no native support.

- CV_FP16: supports load/store into/from float32
- CV_SIMD_FP16: declares FP16 types and native FP16 operations

It is designed for 1D vectors only

keep the same type for passed Scalar values