Metal: Compute shader (Conv.Conv2DKernelKxK_T16x16_R4x4_NHWC) has 256 threads in a threadgroup which is more than the supported max of 224 in this device

Reproduction steps:
1. Download and open this project: https://github.com/cj-mills/Barracuda-PoseNet-Tutorial.
2. Build and run the "SampleScene" to a iPad 6 iOS device (Apple A10 Fusion).
3. Notice the message in the console.

Actual result:
- Metal: Compute shader (Conv.Conv2DKernelKxK_T16x16_R4x4_NHWC) has 256 threads in a threadgroup which is more than the supported max of 224 in this device

Reproduced with:
- iPad 6 (iOS 14.3)
- 1.0.4, 1.3.0(user)

Not reproducible with:
- iPhone X (iOS 11.3.1)

Note;
- The issue doesn't break execution and is cosmetic only.
- May be related to more devices than iPad 6.