feat: Add pagination and field stripping for list_applications

[csz-akuity]

Intended to reduce to amount of tokens used by the list_applications tool.
Related to #59

Add Pagination with Limits/Offsets:

# Example prompts 
List 5 applications to get a quick overview
Search for "frontend" apps, limit to 3 results

Strip manifest fields:

const strippedItems = body.items?.map((app) => ({
  metadata: {
    name: app.metadata?.name,              // ✅ Keep
    namespace: app.metadata?.namespace,    // ✅ Keep
    labels: app.metadata?.labels,          // ✅ Keep
    creationTimestamp: app.metadata?.creationTimestamp  // ✅ Keep
    // ❌ managedFields - REMOVED (1000-3000 tokens)
    // ❌ resourceVersion - REMOVED
    // ❌ uid - REMOVED
    // ❌ ownerReferences - REMOVED
    // ❌ finalizers - REMOVED
  },
  spec: {
    project: app.spec?.project,      // ✅ Keep
    source: app.spec?.source,        // ✅ Keep
    destination: app.spec?.destination  // ✅ Keep
    // ❌ syncPolicy - REMOVED
    // ❌ revisionHistoryLimit - REMOVED
    // ❌ ignoreDifferences - REMOVED
  },
  status: {
    sync: app.status?.sync,      // ✅ Keep
    health: app.status?.health,  // ✅ Keep
    summary: app.status?.summary // ✅ Keep
    // ❌ history - REMOVED (100+ entries)
    // ❌ conditions - REMOVED
    // ❌ resources - REMOVED (full manifests!)
    // ❌ operationState - REMOVED
  }
})) ?? [];

[alexmt]

LGTM. Thank you @csz-akuity !

FYI, we could reduce amount of traffic between mcp server and argocd as well using fields query parameter. E.g. following request should return only resourceVersion, metadata.name, metadata.annotations, and metadata.labels fields:

/api/v1/applications?fields=metadata.resourceVersion,items.metadata.name,items.metadata.namespace,items.metadata.annotations,items.metadata.labels

[alexmt]

Lint failed. Please run pnpm lint --fix to fix linter errors.