feat: add model tiering for agents based on task complexity

Assign models by cognitive demand:
- Opus (inherit): planning, spec, architecture, review agents
- Sonnet: task-group-implementer, test-suite-runner
- Haiku: task-classifier, project-analyzer

Author: mariuszs

plugins/maister/agents/task-classifier.md

@@ -1,7 +1,7 @@
 ---
 name: task-classifier
 description: Task classification specialist analyzing task descriptions and issue references to classify into 5 workflow types (development, performance, migration, research). Supports GitHub/Jira integration, codebase context analysis, and confidence scoring.
-model: inherit
+model: haiku
 color: purple
 ---
 

plugins/maister/agents/task-group-implementer.md

@@ -1,7 +1,7 @@
 ---
 name: task-group-implementer
 description: Execute a single task group from an implementation plan with continuous standards discovery. Writes code, runs tests, returns structured execution report. Does NOT mark checkboxes - main agent handles progress tracking.
-model: inherit
+model: sonnet
 color: green
 ---
 

plugins/maister/agents/test-suite-runner.md

@@ -1,7 +1,7 @@
 ---
 name: test-suite-runner
 description: Runs the full test suite and analyzes results. Identifies test command from project config, executes all tests (not just feature tests), reports pass/fail counts, flags regressions in unrelated areas, and categorizes failures. Read-only - reports issues without fixing. Does not interact with users.
-model: inherit
+model: sonnet
 color: red
 ---