Summary

Adds facets to IssueConnection and AlertConnection, giving clients distribution counts to build filter UIs without extra round-trips.

Changes

Issues — SQL-aggregate pattern

IssueConnection gains an optional facets field:

type IssueConnection {
  pageInfo: PageInfo!
  nodes: [Issue!]!
  edges: [IssueEdge!]!
  facets: IssueFacets
}

type IssueFacets {
  environments:  [StringFacetItem!]!
  severities:    [IssueSeverityFacetItem!]!
  resourceTypes: [IssueResourceTypeFacetItem!]!
  issueTypes:    [IssueTypeFacetItem!]!
}

New supporting types: IssueSeverityFacetItem, IssueResourceTypeFacetItem, IssueTypeFacetItem.

• SQL – new FacetsForIssues aggregate query groups by (severity, resource_type, env, issue_type) and returns filtered_count per group. The scope params (scope_resource_type, scope_resource_name, scope_env) are applied in the outer WHERE so only groups that exist within the connection scope appear in the result; user filter params (env[], severity, issue_type, filter_resource_type, filter_resource_name) go inside COUNT(*) FILTER (...). Single extra DB query, executed only when facets is requested.
• IssueConnection – changed from a type alias to a struct (like ActivityLogEntryConnection) carrying teamSlug, scope, and filter for lazy facet computation in the resolver.
• IssueScope – separates the fixed resource context (set by the caller, e.g. application.issues fixes resourceType, resourceName, env) from the user-controlled IssueFilter. Scope params narrow the base dataset; filter params narrow the counted set.
• ComputeFacets – sums filteredCount per dimension; dimension values with filteredCount = 0 are still present because the scope WHERE ensures every group that exists in the scope seeds the map.
• All six Issues resolvers (team + five resource-level) updated.

Alerts — in-memory pattern

AlertConnection gains an optional facets field:

type AlertConnection {
  pageInfo: PageInfo!
  nodes: [Alert!]!
  edges: [AlertEdge!]!
  facets: AlertFacets
}

type AlertFacets {
  environments: [StringFacetItem!]!
  states:       [AlertStateFacetItem!]!
}

New supporting type: AlertStateFacetItem.

• AlertConnection – changed from pagination.Connection to pagination.FacetableConnection, following the same pattern as applications, OpenSearch, and Valkey.
• AlertFacets – computes environment and state distribution counts in-memory over all alerts in the connection. Uses sync.Once to avoid recomputing the filtered set.
• Both Alerts resolvers (teamResolver, teamEnvironmentResolver) now use SortFilter.PaginatedList.

Tests

Issues — TestBuildFacets covers three cases:

1. Mixed rows with seeded zero-count values — correct aggregation
2. With filter — unmatched dimension values present with count 0
3. Empty input — empty slices, no panics

Alerts — TestAlertFacets_Environments and TestAlertFacets_States cover:

1. No filter — all counts match totals
2. Filter by a dimension — non-matching values present with count 0
3. Filter across dimensions — counts reflect intersection
4. Empty alerts — no facet items

Design choices

• SQL-aggregate for issues, in-memory for alerts — issues live in the DB and can be numerous; alerts are fetched in full from an external source and already in memory, so in-memory counting is simpler and equally efficient.
• Scope vs filter split — scope params narrow the base dataset (so only groups that exist in the connection scope appear), filter params narrow the counted set. This follows the activitylog pattern and ensures facet items are never missing due to the current filter.
• resourceName is not a facet dimension — it is a free-text search field, not a categorical one.