feat: add core experimentation service for startup experiments

[bhavyaus]

Core Experimentation Service for Startup Layout Experiments

This PR introduces a new CoreExperimentationService to manage A/B testing experiments for VS Code startup experiences.

How the Service Works (Pseudocode)

1. On service initialization:
   IF user's first session was more than 1 day ago:
     RETURN (not eligible for startup experiments)
   
   IF experiment already exists in storage:
     RETURN (user already participated)
   
   2. Generate random cohort (0-1)
   
   3. Check if user falls within target percentage:
      - Stable builds: 20% participation  
      - Insider builds: 20% participation
   
   4. IF user is selected:
      - Normalize cohort to experiment range
      - Assign to experiment group based on normalized value:
        * [0-25%): Control group
        * [25-50%): Maximized Chat
        * [50-75%): Split Empty Editor Chat  
        * [75-100%): Split Welcome Chat
      
      - Store experiment data in application storage
      - Set context key for UI consumption
      - Send telemetry event

Service Usage

// Get current user's experiment (returns undefined if not in experiment)
const exp:IExperiment = this.coreExperimentationService.getExperiment();

if (exp?.experimentGroup) {
// Valid values (see StartupExperimentGroup enum): 'control', 'maximizedChat', 'splitEmptyEditorChat', 'splitWelcomeChat'
}

Context Key Usage

The service sets a context key:

// Context key: 'coreExperimentation.startupExpGroup'
// Valid values (see StartupExperimentGroup enum): 'control', 'maximizedChat', 'splitEmptyEditorChat', 'splitWelcomeChat'

Experiment Configuration

• Stable builds: 20% participation, 4 equal groups (25% each)
• Insider builds: 20% participation, 4 equal groups (25% each)

[bpasero]

@bhavyaus thanks, lets get this in in its current state. I think its a bit weird that it would not be generic but include a hardcoded set of startup experiments, likely making it harder if we ever want to use this for other experiments. But we can talk about that when we are there.