feat: add Vertex AI provider for Google Cloud AI/ML integration

[duckyduckycode]

Summary

Adds a new Vertex AI provider that integrates Keep with Google Cloud Vertex AI, resolving #6087.

Vertex AI is Google Cloud's unified ML platform providing access to foundation models (Gemini, etc.) for text generation, structured output, and AI-powered workflows within Keep.

What Changed

New Provider: keep/providers/vertexai_provider/

Dual Authentication Support:

• API Key mode — Simple setup using Google AI Studio / Vertex AI express mode API key
• Service Account mode — Full Vertex AI with service account JSON + project ID + region selection, supporting GKE workload identity

Features:

• _query() method following the same pattern as Anthropic/Gemini/OpenAI providers
• Configurable model selection (default: gemini-2.0-flash, supports all Gemini models)
• Structured output via JSON schema (system instruction injection)
• Automatic JSON response parsing
• Temperature and max_tokens parameters
• All 11 major Vertex AI regions
• Comprehensive error handling with clear error messages

Files Added

• keep/providers/vertexai_provider/__init__.py
• keep/providers/vertexai_provider/vertexai_provider.py
• tests/providers/vertexai_provider/test_vertexai_provider.py
• docs/providers/documentation/vertexai-provider.mdx

Usage in Workflows

actions:
  - name: classify-alert
    provider:
      type: vertexai
      config: "{{ providers.vertexai }}"
    with:
      prompt: "Classify this alert severity: {{ alert.message }}"
      model: gemini-2.0-flash
      structured_output_format:
        type: json_schema
        json_schema:
          name: severity_classification
          schema:
            type: object
            properties:
              severity:
                type: string
                enum: ["critical", "high", "medium", "low", "info"]
            required: ["severity"]
            additionalProperties: false
          strict: true

Testing

15 unit tests covering:

• Auth config validation (API key, service account, missing auth, missing project)
• Provider initialization and type extraction
• Basic query functionality
• JSON auto-parsing
• Plain text preservation
• Structured output with system instruction
• Model selection
• Temperature parameter
• Default model verification
• dispose() lifecycle

Checklist

• Provider follows existing naming convention (*_provider)
• __init__.py exports provider class and auth config
• validate_config() implemented
• dispose() implemented
• PROVIDER_DISPLAY_NAME, PROVIDER_CATEGORY, PROVIDER_TAGS set
• Documentation added
• Unit tests added
• No breaking changes to existing providers

Resolves #6087

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}