Feature: Add macOS MPS support

[ldspnbf]

Closes #68

This PR adds support for PyTorch MPS (Metal Performance Shaders) acceleration on macOS devices.

Changes included:

• Upgraded PyTorch dependency in requirements.txt and pyproject.toml to support MPS operations.
• Modified server/nninteractive_slicer_server/main.py to automatically detect and use torch.device("mps") when available.

This allows Mac users to utilize their GPU for inference.

[coendevente]

Thank you for your contribution. I think it'd be great to have the server work on macOS.

I tried out your code on my MacBook (M1 Pro) and it doesn't seem to work, though. When I try to build using build.sh, I get the following error:

[+] Building 4.5s (6/18)                                                                                                                                   
 => [internal] load build definition from Dockerfile                                                                                                  0.0s
 => => transferring dockerfile: 37B                                                                                                                   0.0s
 => [internal] load .dockerignore                                                                                                                     0.0s
 => => transferring context: 2B                                                                                                                       0.0s
 => [internal] load metadata for docker.io/library/ubuntu:20.04                                                                                       0.4s
 => CACHED [ 1/14] FROM docker.io/library/ubuntu:20.04@sha256:8feb4d8ca5354def3d8fce243717141ce31e2c428701f6682bd2fafe15388214                        0.0s
 => [internal] load build context                                                                                                                     0.0s
 => => transferring context: 262B                                                                                                                     0.0s
 => ERROR [ 2/14] RUN apt-get update && apt-get install -y wget python3.12 python3.12-pip                                                             4.0s
------                                                                                                                                                     
 > [ 2/14] RUN apt-get update && apt-get install -y wget python3.12 python3.12-pip:                                                                        
#5 0.416 Get:1 http://ports.ubuntu.com/ubuntu-ports focal InRelease [265 kB]                                                                               
#5 0.957 Get:2 http://ports.ubuntu.com/ubuntu-ports focal-updates InRelease [128 kB]                                                                       
#5 1.091 Get:3 http://ports.ubuntu.com/ubuntu-ports focal-backports InRelease [128 kB]                                                                     
#5 1.219 Get:4 http://ports.ubuntu.com/ubuntu-ports focal-security InRelease [128 kB]                                                                      
#5 1.350 Get:5 http://ports.ubuntu.com/ubuntu-ports focal/multiverse arm64 Packages [139 kB]
#5 1.393 Get:6 http://ports.ubuntu.com/ubuntu-ports focal/restricted arm64 Packages [1317 B]
#5 1.393 Get:7 http://ports.ubuntu.com/ubuntu-ports focal/main arm64 Packages [1234 kB]
#5 1.614 Get:8 http://ports.ubuntu.com/ubuntu-ports focal/universe arm64 Packages [11.1 MB]
#5 2.092 Get:9 http://ports.ubuntu.com/ubuntu-ports focal-updates/main arm64 Packages [3822 kB]
#5 2.293 Get:10 http://ports.ubuntu.com/ubuntu-ports focal-updates/universe arm64 Packages [1505 kB]
#5 2.330 Get:11 http://ports.ubuntu.com/ubuntu-ports focal-updates/restricted arm64 Packages [82.9 kB]
#5 2.331 Get:12 http://ports.ubuntu.com/ubuntu-ports focal-updates/multiverse arm64 Packages [14.9 kB]
#5 2.331 Get:13 http://ports.ubuntu.com/ubuntu-ports focal-backports/universe arm64 Packages [27.8 kB]
#5 2.332 Get:14 http://ports.ubuntu.com/ubuntu-ports focal-backports/main arm64 Packages [54.8 kB]
#5 2.333 Get:15 http://ports.ubuntu.com/ubuntu-ports focal-security/restricted arm64 Packages [77.0 kB]
#5 2.338 Get:16 http://ports.ubuntu.com/ubuntu-ports focal-security/universe arm64 Packages [1214 kB]
#5 2.361 Get:17 http://ports.ubuntu.com/ubuntu-ports focal-security/main arm64 Packages [3425 kB]
#5 2.449 Get:18 http://ports.ubuntu.com/ubuntu-ports focal-security/multiverse arm64 Packages [8083 B]
#5 2.805 Fetched 23.4 MB in 3s (9032 kB/s)
#5 2.805 Reading package lists...
#5 3.338 Reading package lists...
#5 3.803 Building dependency tree...
#5 3.890 Reading state information...
#5 3.943 E: Unable to locate package python3.12-pip
#5 3.943 E: Couldn't find any package by glob 'python3.12-pip'
#5 3.943 E: Couldn't find any package by regex 'python3.12-pip'
------
executor failed running [/bin/sh -c apt-get update && apt-get install -y wget python3.12 python3.12-pip]: exit code: 100

Does the build.sh actually work in your case on a macOS machine?

The fact that you use a different base image seems to cause issues. Furthermore, since you don't use a base image with cuda installed already anymore, I expect this will fail on machines with CUDA supported GPUs, now. Haven't tested this yet, though.

Also, I spotted another issue in your PR. You update the torch version to 2.9.0. However, the nnInteractive authors explicitly instruct not to use this version:

⚠️ Temporary Compatibility Warning
There is a known issue with PyTorch 2.9.0 causing OOM errors during inference in nnInteractive (related to 3D convolutions — see the PyTorch issue pytorch/pytorch#166122).
Until this is resolved, please use PyTorch 2.8.0 or earlier.