fix(drone-dep): revert _count_deploy=False — dep deploys must count for DG4.1

The DG4.1 formula in run_recipe_ci.py is:
  expected_deploy_count = 1 + deps_deployed_count

So when gitea dep deploys, the expected count becomes 2 (1 recipe + 1 dep).
The _count_deploy=False fix made dep deploys NOT count, giving actual=1 vs
expected=2 → DG4.1 violation even though the run was correct.

Original error "deploy-count 2 != 1" was because deps_state was empty when
the DG4.1 check ran (provisioning had failed), giving expected=1 while count
was already 2 from an early dep deploy. The proper fix is for _provision_deps
to succeed (which it now does), not to suppress counting.

Revert _count_deploy=False in deps.py; update docstrings for clarity.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

runner/harness/deps.py

@@ -16,6 +16,8 @@ Per Phase-2 DECISIONS:
   must share the single node's MAX_TESTS budget without exceeding it).
 - Each dep is undeployed in the orchestrator's `finally`, in **reverse** order so a recipe-under-
   test can depend on multiple deps with a dependency chain (a → b → c teardown is c → b → a).
+- Dep deploys DO count toward the DG4.1 deploy-count invariant. The formula in run_recipe_ci.py is
+  `expected_deploy_count = 1 + deps_deployed_count`, so each dep deploy increments the counter.
 
 Run state:
 - `$CCCI_DEPS_FILE` — JSON file written by the orchestrator after each dep deploys; each entry is
@@ -80,10 +82,9 @@ def deploy_deps(
     for dep in deps:
         domain = dep_domain(parent_recipe, pr, ref, dep)
         print(f"  dep: deploying {dep} -> {domain}", flush=True)
-        # Dep deploys do NOT count toward the DG4.1 "one deploy per run" invariant — that
+        # Dep deploys count toward DG4.1 — the check expects (1 + len(cold-deps)), so each
-        # contract covers the recipe-under-test only; each dep is a supporting service, not the
+        # dep that deploys here MUST be counted. The formula is authoritative in run_recipe_ci.py:
-        # subject of the test. Pass _count_deploy=False so the main recipe's single-deploy
+        #   expected_deploy_count = 1 + deps_deployed_count
-        # assertion isn't distorted by the number of deps declared.
         dm = meta_for.get(dep) or meta_mod.load(dep)
         lifecycle.deploy_app(
             dep,
@@ -91,7 +92,6 @@ def deploy_deps(
             secrets=True,
             deploy_timeout=int(dm.DEPLOY_TIMEOUT),
             meta=dm,
-            _count_deploy=False,
         )
         try:
             lifecycle.wait_healthy(

runner/harness/lifecycle.py

@@ -254,8 +254,10 @@ def deploy_app(
     past the 900s default. abra's INTERNAL TIMEOUT (recipe's TIMEOUT env, default 300s) is set via
     EXTRA_ENV; this is the Python subprocess wrapper's timeout so abra doesn't get SIGKILLed mid-deploy.
 
-    `_count_deploy`: set False for dep deployments — the DG4.1 "one deploy per run" invariant
+    `_count_deploy`: internal escape hatch — set False to skip incrementing the DG4.1 deploy
-    counts ONLY the recipe-under-test, not its install-time deps (deps_mod.deploy_deps)."""
+    counter (e.g. for test fixtures that call deploy_app without participating in a real run).
+    Normal orchestration should always use the default True — dep deploys count too (the DG4.1
+    formula is `expected = 1 + deps_count`, so deps MUST be counted; see run_recipe_ci.py)."""
     if meta is None:
         meta = meta_mod.load(recipe)
     if _count_deploy: