fix(tests): align gen-test-tasks assertions with v4.1 calc-tasks floor

Follow-up to bb269b5 (context-aware task count formula). v4.1 dropped the
calc-tasks floor from 10 to 3 and added the [3, 25] clamp, which broke two
test assumptions:

- test_user_e2e: replace `>= 10` with `3 <= recommended <= 25`
- both files: `gen-test-tasks` needs >=5 tasks to emit a USER-TEST
  checkpoint; use max(5, recommended) so small PRDs still exercise the
  checkpoint path.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: claude

tests/test_edge_cases.py

@@ -296,9 +296,12 @@ def test_full_workflow_preflight_to_validate(self, tmp_project, sample_prd, samp
         rc, calc = run_script(SCRIPT_PY, ["calc-tasks", "--requirements", "6"])
         assert rc == 0
 
-        # Step 4: Generate test tasks
+        # Step 4: Generate test tasks — gen-test-tasks inserts a checkpoint
+        # every 5 tasks, so we need >= 5. v4.1 calc-tasks may recommend
+        # fewer than 5 for small PRDs (floor is 3, not 10), so use max.
+        total_for_checkpoints = max(5, calc["recommended"])
         rc, test_tasks = run_script(SCRIPT_PY, [
-            "gen-test-tasks", "--total", str(calc["recommended"])
+            "gen-test-tasks", "--total", str(total_for_checkpoints)
         ])
         assert rc == 0
         assert test_tasks["test_tasks_generated"] > 0

tests/test_user_e2e.py

@@ -84,11 +84,17 @@ def test_step3_full_workflow_end_to_end(self, user_project):
         num_reqs = int(num_match.group(1)) if num_match else 5
         rc, calc = run_script(SCRIPT_PY, ["calc-tasks", "--requirements", str(num_reqs)])
         assert rc == 0
-        assert calc["recommended"] >= 10
-
-        # Step 9: Generate test checkpoints
+        # v4.1 calc-tasks floor is 3 (was 10). Recommended is context-aware
+        # and clamped to [3, 25]. For an e2e test that doesn't pin team/scope,
+        # assert the value is within the clamp window.
+        assert 3 <= calc["recommended"] <= 25
+
+        # gen-test-tasks needs at least 5 tasks to produce a checkpoint
+        # (inserts a USER-TEST every 5 tasks). Use max(5, recommended) so
+        # the downstream step still has something to assert on.
+        total_for_checkpoints = max(5, calc["recommended"])
         rc, checkpoints = run_script(SCRIPT_PY, [
-            "gen-test-tasks", "--total", str(calc["recommended"])
+            "gen-test-tasks", "--total", str(total_for_checkpoints)
         ])
         assert rc == 0
         assert checkpoints["test_tasks_generated"] >= 1