agents: multi-step agent support

Split agent prompts on === PROMPT === delimiter. Each step runs as
a new user message in the same LLM conversation, so context carries
forward naturally between steps. Single-step agents are unchanged.

- AgentDef.prompt -> AgentDef.prompts: Vec<String>
- AgentBatch.prompt -> AgentBatch.prompts: Vec<String>
- API layer injects next prompt after each text response
- {{conversation:N}} parameterized byte budget for conversation context

Co-Authored-By: Kent Overstreet <kent.overstreet@linux.dev>

src/subconscious/digest.rs

@@ -226,7 +226,7 @@ fn generate_digest(
     // Load prompt from agent file; fall back to prompts dir
     let def = super::defs::get_def("digest");
     let template = match &def {
-        Some(d) => d.prompt.clone(),
+        Some(d) => d.prompts.first().cloned().unwrap_or_default(),
         None => {
             let path = crate::config::get().prompts_dir.join("digest.md");
             std::fs::read_to_string(&path)