[BUG] channel=1 is not supported in resize_crop_convert_reformat.

[caffeinism]

Describe the bug
The document's C API states the following

Limitations:
Input:
Data Layout: [NVCV_TENSOR_HWC, NVCV_TENSOR_NHWC]
Channels: [1, 3]

However, channel=1 is not supported.

Steps/Code to reproduce bug

import torch
import cvcuda

src = cvcuda.as_tensor(torch.zeros((1080, 1920, 1), dtype=torch.uint8).cuda(), layout="HWC")
cvcuda.resize_crop_convert_reformat(src=src, resize_dim=(1280, 720), interp=cvcuda.Interp.LINEAR, crop_rect=cvcuda.RectI(y=100, x=200, width=300, height=400), layout="HWC", data_type=cvcuda.Type.U8)

NVCV_ERROR_NOT_COMPATIBLE: Only three-channel input is currently supported: Provided 1 input channels

Environment overview (please complete the following information)

• Environment location: docker
• Method of cuDF install: pip