Add comprehensive Nebius provider README

- Document supported features based on Nebius API research
- Include instance management, GPU clusters, images, and quotas
- List unsupported features with explanations
- Follow Lambda Labs README structure for consistency
- Reference official Nebius API documentation and Go SDK

Co-Authored-By: Alec Fong <alecsanf@usc.edu>

Author: devin-ai-integration[bot]

internal/nebius/v1/README.md

@@ -0,0 +1,89 @@
+# Nebius Provider
+
+This directory contains the Nebius provider implementation for the compute package.
+
+## Overview
+
+The Nebius provider implements the CloudClient interface defined in `pkg/v1` to provide access to Nebius AI Cloud infrastructure. This implementation is based on the official Nebius API documentation at https://github.com/nebius/api and uses the Nebius Go SDK.
+
+## Supported Features
+
+Based on the Nebius API documentation, the following features are **SUPPORTED**:
+
+### Instance Management
+- ✅ **Create Instance**: `InstanceService.Create` in compute/v1/instance_service.proto
+- ✅ **Get Instance**: `InstanceService.Get` and `InstanceService.GetByName` 
+- ✅ **List Instances**: `InstanceService.List` with pagination support
+- ✅ **Terminate Instance**: `InstanceService.Delete`
+- ✅ **Stop Instance**: `InstanceService.Stop`
+- ✅ **Start Instance**: `InstanceService.Start`
+- ✅ **Update Instance**: `InstanceService.Update`
+
+### GPU Cluster Management
+- ✅ **Create GPU Cluster**: `GpuClusterService.Create` in compute/v1/gpu_cluster_service.proto
+- ✅ **Get GPU Cluster**: `GpuClusterService.Get` and `GpuClusterService.GetByName`
+- ✅ **List GPU Clusters**: `GpuClusterService.List` with pagination support
+- ✅ **Delete GPU Cluster**: `GpuClusterService.Delete`
+- ✅ **Update GPU Cluster**: `GpuClusterService.Update`
+
+### Machine Images
+- ✅ **Get Images**: `ImageService.Get`, `ImageService.GetByName`, `ImageService.GetLatestByFamily`
+- ✅ **List Images**: `ImageService.List` with filtering support
+
+### Quota Management
+- ✅ **Get Quotas**: `QuotaAllowanceService` in quotas/v1/quota_allowance_service.proto
+
+## Unsupported Features
+
+The following features are **NOT SUPPORTED** (no clear API endpoints found):
+
+### Instance Operations
+- ❌ **Reboot Instance**: No reboot endpoint found in instance_service.proto
+- ❌ **Instance Tags**: No dedicated tagging service found
+- ❌ **Change Instance Type**: No instance type modification endpoint
+
+### Volume Management
+- ❌ **Resize Instance Volume**: Volume resizing not clearly documented
+
+### Location Management
+- ❌ **Get Locations**: No location listing service found
+
+### Firewall Management
+- ❌ **Firewall Rules**: Network security handled through VPC service, not instance-level firewall rules
+
+## Implementation Approach
+
+This implementation uses the `NotImplCloudClient` pattern for unsupported features:
+- Supported features have TODO implementations with API service references
+- Unsupported features return `ErrNotImplemented` (handled by embedded NotImplCloudClient)
+- Full CloudClient interface compliance is maintained
+
+## Nebius API
+
+The provider integrates with the Nebius AI Cloud API:
+- Base URL: `{service-name}.api.nebius.cloud:443` (gRPC)
+- Authentication: Service account based (JWT tokens)
+- SDK: `github.com/nebius/gosdk`
+- Documentation: https://github.com/nebius/api
+- API Type: Locational (location-specific endpoints)
+
+## Key Features
+
+Nebius AI Cloud is known for:
+- GPU instances and GPU clusters for AI/ML workloads
+- Comprehensive compute, storage, and networking services
+- gRPC-based API with strong typing
+- Service account authentication with JWT tokens
+- Location-specific API endpoints
+- Advanced operations tracking and idempotency
+- Integration with VPC, IAM, billing, and quota services
+- Container registry and managed services
+
+## TODO
+
+- [ ] Implement actual API integration for supported features
+- [ ] Add proper service account authentication handling
+- [ ] Add comprehensive error handling and retry logic
+- [ ] Add logging and monitoring
+- [ ] Add comprehensive testing
+- [ ] Investigate VPC integration for networking features