Remove dead action pipeline: parsing, depth tracking, knowledge loop, fact miner

Agents now apply changes via tool calls (poc-memory write/link-add/etc)
during the LLM call. The old pipeline — where agents output WRITE_NODE/
LINK/REFINE text, which was parsed and applied separately — is dead code.

Removed:
- Action/ActionKind/Confidence types and all parse_* functions
- DepthDb, depth tracking, confidence gating
- apply_action, stamp_content, has_edge
- NamingResolution, resolve_naming and related naming agent code
- KnowledgeLoopConfig, CycleResult, GraphMetrics, convergence checking
- run_knowledge_loop, run_cycle, check_convergence
- apply_consolidation (old report re-processing)
- fact_mine.rs (folded into observation agent)
- resolve_action_names

Simplified:
- AgentResult no longer carries actions/no_ops
- run_and_apply_with_log just runs the agent
- consolidate_full simplified action tracking

-1364 lines.

poc-memory/src/agents/daemon.rs

@@ -59,8 +59,8 @@ fn job_consolidation_agent(
             ctx.log_line(msg);
             log_event(&job_name2, "progress", msg);
         };
-        let (total, applied) = super::knowledge::run_and_apply_with_log(&mut store, &agent, batch, "consolidate", &log)?;
-        ctx.log_line(&format!("done: {} actions ({} applied)", total, applied));
+        super::knowledge::run_and_apply_with_log(&mut store, &agent, batch, "consolidate", &log)?;
+        ctx.log_line("done");
         Ok(())
     })
 }
@@ -377,20 +377,10 @@ fn job_digest_links(ctx: &ExecutionContext) -> Result<(), TaskError> {
     })
 }
 
-fn job_knowledge_loop(ctx: &ExecutionContext) -> Result<(), TaskError> {
-    run_job(ctx, "knowledge-loop", || {
-        let config = super::knowledge::KnowledgeLoopConfig {
-            max_cycles: 100,
-            batch_size: 5,
-            ..Default::default()
-        };
-        ctx.log_line("running agents");
-        let results = super::knowledge::run_knowledge_loop(&config)?;
-        ctx.log_line(format!("{} cycles, {} actions",
-            results.len(),
-            results.iter().map(|r| r.total_applied).sum::<usize>()));
-        Ok(())
-    })
+fn job_knowledge_loop(_ctx: &ExecutionContext) -> Result<(), TaskError> {
+    // Knowledge loop removed — agents now use tool calls directly.
+    // Consolidation is handled by consolidate_full() in the consolidate job.
+    Ok(())
 }
 
 fn job_digest(ctx: &ExecutionContext) -> Result<(), TaskError> {