fix(workflows): close review gaps — pull round-trip, dup-names, reclaim budget

Adversarial review of the v1.2 Workflows track surfaced two HIGH footguns and
several correctness/hardening gaps. Fixes:

HIGH
- pull round-trip: `pic pull` dropped `[[workflows]]`, so a later
  `apply --prune` silently deleted them. The list endpoint now returns the full
  `definition`; pull rebuilds the manifest block (inverse of workflow_to_wire).
- case-colliding names: two workflows differing only by case collided in the
  reconcile diff (keyed by lower(name)), silently dropping one. Rejected up
  front in validate_bundle_for.

MEDIUM
- reclaim retry budget: a crashed attempt (no outcome) consumed the retry
  budget. reclaim_stale_steps now decrements `attempt` (floored) and clears
  `next_attempt_at`, so a crash no longer counts as a failed try.
- on_error/backoff: typed the manifest fields against the shared enums so a
  bad value fails at TOML parse with a clear message, not an opaque 500.
- dedupe depends_on: a repeated dependency inflated the Kahn in-degree past the
  single decrement, reporting a spurious cycle for a valid DAG. Count distinct.
- canceled child: a canceled sub-workflow resolved the parent step with a
  message naming the cause instead of a generic "failed"; documented that a
  run-level cancel op is not yet supported.

LOW
- list_run_steps is now app-scoped at the query (JOIN workflow_runs) rather
  than relying on caller pre-verification.
- partial failure is surfaced: a run that succeeds with on_error=continue
  failures records them in the run's `error` field.
- admin-started runs log the root_execution_id against the principal.
- documented the deliberate when(missing→false) vs template(missing→fail)
  asymmetry; corrected the claim's atomicity comment.

Tests: unit (dedupe-deps), DB-gated reclaim-budget assertion, and two new CLI
journeys (duplicate-name rejection, pull→plan clean round-trip). fmt + clippy
-D warnings clean; 450 lib + 14 orchestrator DB + 5 workflow journeys pass;
schema snapshot unchanged (no migration).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

crates/picloud-cli/src/cmds/pull.rs

@@ -18,8 +18,9 @@ use crate::client::{Client, VarOwnerArg};
 use crate::config;
 use crate::manifest::{
     CronTriggerSpec, DocsTriggerSpec, FilesTriggerSpec, KvTriggerSpec, Manifest, ManifestApp,
-    ManifestRoute, ManifestScript, ManifestSecrets, ManifestTriggers, PubsubTriggerSpec,
-    QueueTriggerSpec, MANIFEST_FILE,
+    ManifestRoute, ManifestScript, ManifestSecrets, ManifestTriggers, ManifestWorkflow,
+    ManifestWorkflowRetry, ManifestWorkflowStep, PubsubTriggerSpec, QueueTriggerSpec,
+    MANIFEST_FILE,
 };
 use crate::output::{KvBlock, OutputMode};
 
@@ -45,6 +46,7 @@ pub async fn run(app_ident: &str, dir: &Path, force: bool, mode: OutputMode) ->
     let app = client.apps_get(app_ident).await?;
     let scripts = client.scripts_list_by_app(app_ident).await?;
     let triggers = client.triggers_list(app_ident).await?.triggers;
+    let workflows = client.workflows_list(app_ident).await?;
     let secrets = client
         .secrets_list(VarOwnerArg::App(app_ident), None)
         .await?
@@ -251,7 +253,7 @@ pub async fn run(app_ident: &str, dir: &Path, force: bool, mode: OutputMode) ->
         },
         vars: manifest_vars,
         suppress: crate::manifest::ManifestSuppress::default(),
-        workflows: Vec::new(),
+        workflows: workflows.iter().map(wire_workflow_to_manifest).collect(),
     };
 
     std::fs::write(&manifest_path, manifest.to_toml()?)
@@ -264,7 +266,8 @@ pub async fn run(app_ident: &str, dir: &Path, force: bool, mode: OutputMode) ->
         .field("scripts", manifest.scripts.len().to_string())
         .field("routes", manifest.routes.len().to_string())
         .field("triggers", trigger_count(&manifest.triggers).to_string())
-        .field("secrets", manifest.secrets.names.len().to_string());
+        .field("secrets", manifest.secrets.names.len().to_string())
+        .field("workflows", manifest.workflows.len().to_string());
     block.print(mode);
     Ok(())
 }
@@ -273,6 +276,70 @@ fn trigger_count(t: &ManifestTriggers) -> usize {
     t.kv.len() + t.docs.len() + t.files.len() + t.cron.len() + t.pubsub.len() + t.queue.len()
 }
 
+/// Rebuild a `[[workflows]]` manifest block from the server's stored definition
+/// (the inverse of `plan::workflow_to_wire`). Without this a `pic pull` would
+/// drop the app's workflows, and a subsequent `pic apply --prune` would delete
+/// them from the server.
+fn wire_workflow_to_manifest(w: &crate::client::WorkflowDto) -> ManifestWorkflow {
+    ManifestWorkflow {
+        name: w.name.clone(),
+        enabled: w.enabled,
+        steps: w
+            .definition
+            .steps
+            .iter()
+            .map(wire_step_to_manifest)
+            .collect(),
+    }
+}
+
+fn wire_step_to_manifest(s: &picloud_shared::WorkflowStepDef) -> ManifestWorkflowStep {
+    ManifestWorkflowStep {
+        name: s.name.clone(),
+        function: s.function.clone(),
+        workflow: s.workflow.clone(),
+        // JSON step input → TOML. A whole-null input is the "absent" encoding;
+        // anything else round-trips through serde. A value TOML can't represent
+        // (e.g. a nested null) is dropped with a warning rather than aborting
+        // the whole pull.
+        input: if s.input.is_null() {
+            None
+        } else {
+            match toml::Value::try_from(&s.input) {
+                Ok(tv) => Some(tv),
+                Err(e) => {
+                    eprintln!(
+                        "warning: workflow step `{}` input not representable in TOML ({e}); \
+                         omitting it from the manifest",
+                        s.name
+                    );
+                    None
+                }
+            }
+        },
+        depends_on: s.depends_on.clone(),
+        when: s.when.clone(),
+        retry: s.retry.as_ref().map(|r| ManifestWorkflowRetry {
+            max_attempts: r.max_attempts,
+            // Exponential is the default → omit; likewise the default base_ms.
+            backoff: match r.backoff {
+                picloud_shared::WorkflowBackoff::Exponential => None,
+                other => Some(other),
+            },
+            base_ms: if r.base_ms == 500 {
+                None
+            } else {
+                Some(r.base_ms)
+            },
+        }),
+        // Fail is the default → omit; Continue is explicit.
+        on_error: match s.on_error {
+            picloud_shared::OnError::Fail => None,
+            picloud_shared::OnError::Continue => Some(picloud_shared::OnError::Continue),
+        },
+    }
+}
+
 /// True if `name` is safe to use as a single path component in `scripts/`.
 /// Rejects empty/over-long names, path separators, `.`/`..`, leading dots,
 /// and any deceptive display character — a server-returned name is otherwise