Enable SHADER_F16 feature only when available

[Ericky14]

Recent Shader_F16 broke WASM, this should fix it

[miredirex]

And it also broke WGPU_BACKEND=dx12 cargo run on Windows somehow, even though I'm on a very modern Nvidia GPU. Haven't tested with this PR, just rolled back to commit fd25ff1 for the time being. Attached some logs:

Logs

(...there is more wgpu debug logs above that I decided not to include...)

[DEBUG wgpu_core::instance] Found 3 compatible adapters. Sorted by preference:
[DEBUG wgpu_core::instance] * AdapterInfo { name: "NVIDIA GeForce RTX 4070 Ti SUPER", vendor: 4318, device: 9989, device_type: DiscreteGpu, driver: "32.0.15.8180", driver_info: "", backend: Dx12 }
[DEBUG wgpu_core::instance] * AdapterInfo { name: "Intel(R) UHD Graphics 770", vendor: 32902, device: 42880, device_type: IntegratedGpu, driver: "32.0.101.7040", driver_info: "", backend: Dx12 }
[DEBUG wgpu_core::instance] * AdapterInfo { name: "Microsoft Basic Render Driver", vendor: 5140, device: 140, device_type: Cpu, driver: "10.0.26100.7309", driver_info: "", backend: Dx12 }
[DEBUG wgpu_core::instance] Request adapter result AdapterInfo { name: "NVIDIA GeForce RTX 4070 Ti SUPER", vendor: 4318, device: 9989, device_type: DiscreteGpu, driver: "32.0.15.8180", driver_info: "", backend: Dx12 }
[INFO  iced_wgpu::window::compositor] Selected: AdapterInfo {
        name: "NVIDIA GeForce RTX 4070 Ti SUPER",
        vendor: 4318,
        device: 9989,
        device_type: DiscreteGpu,
        driver: "32.0.15.8180",
        driver_info: "",
        backend: Dx12,
    }
[INFO  iced_wgpu::window::compositor] Available formats: Copied {
        it: Iter(
            [
                Bgra8UnormSrgb,
                Rgba8UnormSrgb,
                Bgra8Unorm,
                Rgba8Unorm,
                Rgb10a2Unorm,
                Rgba16Float,
            ],
        ),
    }
[INFO  iced_wgpu::window::compositor] Available alpha modes: [
        Opaque,
    ]
[INFO  iced_wgpu::window::compositor] Selected format: Bgra8Unorm with alpha mode: Auto
Error: GraphicsCreationFailed(GraphicsAdapterNotFound { backend: "wgpu", reason: RequestFailed("no device request succeeded: [(Limits { max_texture_dimension_1d: 8192, max_texture_dimension_2d: 8192, max_texture_dimension_3d: 2048, max_texture_array_layers: 256, max_bind_groups: 2, max_bindings_per_bind_group: 1000, max_dynamic_uniform_buffers_per_pipeline_layout: 8, max_dynamic_storage_buffers_per_pipeline_layout: 4, max_sampled_textures_per_shader_stage: 16, max_samplers_per_shader_stage: 16, max_storage_buffers_per_shader_stage: 8, max_storage_textures_per_shader_stage: 4, max_uniform_buffers_per_shader_stage: 12, max_binding_array_elements_per_shader_stage: 0, max_binding_array_sampler_elements_per_shader_stage: 0, max_uniform_buffer_binding_size: 65536, max_storage_buffer_binding_size: 134217728, max_vertex_buffers: 8, max_buffer_size: 268435456, max_vertex_attributes: 16, max_vertex_buffer_array_stride: 2048, min_uniform_buffer_offset_alignment: 256, min_storage_buffer_offset_alignment: 256, max_inter_stage_shader_components: 60, max_color_attachments: 8, max_color_attachment_bytes_per_sample: 32, max_compute_workgroup_storage_size: 16384, max_compute_invocations_per_workgroup: 256, max_compute_workgroup_size_x: 256, max_compute_workgroup_size_y: 256, max_compute_workgroup_size_z: 64, max_compute_workgroups_per_dimension: 65535, min_subgroup_size: 0, max_subgroup_size: 0, max_push_constant_size: 0, max_non_sampler_bindings: 2048, max_task_workgroup_total_count: 0, max_task_workgroups_per_dimension: 0, max_mesh_output_layers: 0, max_mesh_multiview_count: 0, max_blas_primitive_count: 0, max_blas_geometry_count: 0, max_tlas_instance_count: 0, max_acceleration_structures_per_shader_stage: 0 }, RequestDeviceError { inner: Core(UnsupportedFeature(Features { features_wgpu: FeaturesWGPU(0x0), features_webgpu: FeaturesWebGPU(SHADER_F16) })) }), (Limits { max_texture_dimension_1d: 2048, max_texture_dimension_2d: 2048, max_texture_dimension_3d: 256, max_texture_array_layers: 256, max_bind_groups: 2, max_bindings_per_bind_group: 1000, max_dynamic_uniform_buffers_per_pipeline_layout: 8, max_dynamic_storage_buffers_per_pipeline_layout: 4, max_sampled_textures_per_shader_stage: 16, max_samplers_per_shader_stage: 16, max_storage_buffers_per_shader_stage: 4, max_storage_textures_per_shader_stage: 4, max_uniform_buffers_per_shader_stage: 12, max_binding_array_elements_per_shader_stage: 0, max_binding_array_sampler_elements_per_shader_stage: 0, max_uniform_buffer_binding_size: 16384, max_storage_buffer_binding_size: 134217728, max_vertex_buffers: 8, max_buffer_size: 268435456, max_vertex_attributes: 16, max_vertex_buffer_array_stride: 2048, min_uniform_buffer_offset_alignment: 256, min_storage_buffer_offset_alignment: 256, max_inter_stage_shader_components: 60, max_color_attachments: 4, max_color_attachment_bytes_per_sample: 32, max_compute_workgroup_storage_size: 16352, max_compute_invocations_per_workgroup: 256, max_compute_workgroup_size_x: 256, max_compute_workgroup_size_y: 256, max_compute_workgroup_size_z: 64, max_compute_workgroups_per_dimension: 65535, min_subgroup_size: 0, max_subgroup_size: 0, max_push_constant_size: 0, max_non_sampler_bindings: 2048, max_task_workgroup_total_count: 0, max_task_workgroups_per_dimension: 0, max_mesh_output_layers: 0, max_mesh_multiview_count: 0, max_blas_primitive_count: 0, max_blas_geometry_count: 0, max_tlas_instance_count: 0, max_acceleration_structures_per_shader_stage: 0 }, RequestDeviceError { inner: Core(UnsupportedFeature(Features { features_wgpu: FeaturesWGPU(0x0), features_webgpu: FeaturesWebGPU(SHADER_F16) })) })]") })
error: process didn't exit successfully: `target\debug\program.exe` (exit code: 1)

[badosu]

Maybe related: squidowl/halloy#1411 (comment)

Edit: I tested and confirmed that the PR fixes the issue above for me. Polaris architecture amd gpus should be affected as well (mine is RX580).