First pass at getting layouts

[ahuang11]

First pass at implementing #1447

layout.mp4

The Three-Step Approach:
Break layout generation into three sequential steps similar to Planner -> Agents:
Planning (1 LLM call): Create a layout plan with natural language instructions for each subplot
Parallel Generation (N LLM calls): Generate each subplot independently in parallel
Assembly (Python): Programmatically combine specs into hconcat/vconcat structure

Note, these are the valid properties outside:

spec:
  $schema: https://vega.github.io/schema/vega-lite/v5.json
  
  # SHARED by all charts
  data:
    name: token_use_and_response_time_from_gpt_4o_mini_csv
  
  # Optional: overall title for entire visualization
  title:
    text: GPT-4o Mini Token Usage Analysis
  
  # Optional: shared transforms applied before splitting
  transform:
    - filter: "datum.response_time > 0"
  
  # The concatenation container
  hconcat: [...]

Within each subplot:

hconcat:
- # FIRST CHART
  width: 300                    # This chart's width
  height: container             # This chart's height
  title: Input Tokens Distribution  # This chart's title
  mark: bar                     # This chart's mark type
  encoding:                     # This chart's encodings
    x: {...}
    y: {...}
  transform:                    # Optional: transforms just for this chart
    - filter: "datum.input_tokens < 10000"

- # SECOND CHART  
  width: 300                    # Different width possible
  height: container
  title: Output Tokens Distribution
  mark: point                   # Different mark type
  encoding:
    x: {...}
    y: {...}

Visual Representation

┌─────────────────────────────────────────────────────┐
│ spec (top-level)                                    │
│                                                     │
│  ┌─────────────────────────────────────────────┐  │
│  │ data: (SHARED by all)                       │  │
│  └─────────────────────────────────────────────┘  │
│                                                     │
│  ┌─────────────────────────────────────────────┐  │
│  │ hconcat: [                                  │  │
│  │   ┌──────────────────────────┐              │  │
│  │   │ Chart 1                  │              │  │
│  │   │ - width                  │              │  │
│  │   │ - height                 │              │  │
│  │   │ - title                  │              │  │
│  │   │ - mark                   │              │  │
│  │   │ - encoding               │              │  │
│  │   └──────────────────────────┘              │  │
│  │   ┌──────────────────────────┐              │  │
│  │   │ Chart 2                  │              │  │
│  │   │ - width                  │              │  │
│  │   │ - height                 │              │  │
│  │   │ - title                  │              │  │
│  │   │ - mark                   │              │  │
│  │   │ - encoding               │              │  │
│  │   └──────────────────────────┘              │  │
│  │ ]                                           │  │
│  └─────────────────────────────────────────────┘  │
└─────────────────────────────────────────────────────┘

TODOs:

• make it more token efficient
• fix layout height / width (if both are containers, it gets clipped)

Would also like to address #1431

[codecov]

Codecov Report

❌ Patch coverage is 37.80864% with 403 lines in your changes missing coverage. Please review.
✅ Project coverage is 49.93%. Comparing base (4855124) to head (d64529d).

Files with missing lines	Patch %	Lines
lumen/ai/agents.py	5.59%	152 Missing ⚠️
lumen/tests/views/test_base.py	24.32%	112 Missing ⚠️
lumen/tests/ai/test.py	11.25%	71 Missing ⚠️
lumen/tests/ai/conftest.py	24.07%	41 Missing ⚠️
lumen/views/base.py	4.16%	23 Missing ⚠️
lumen/ai/utils.py	89.18%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1453      +/-   ##
==========================================
+ Coverage   47.89%   49.93%   +2.03%     
==========================================
  Files         122      124       +2     
  Lines       20802    21252     +450     
==========================================
+ Hits         9964    10613     +649     
+ Misses      10838    10639     -199     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[ahuang11]

Area.mp4

[ahuang11]

Maybe remove