Agent-built designs: stop redoing whole-design work on every command (#1148)

A design built over MCP is many small commands, and each one did work over the
whole design, so building or showing a 1,300-body design took minutes:

- moved components were transformed and meshed again after every change; world_state
  now reuses each placed body whose definition and placement are unchanged
- next_d_name searched the design once per candidate number
- Document::eval evaluated every parameter for every expression; now only when the
  expression names one
- param_rows' used_by was quadratic; the desktop no longer gathers completions on every
  change while no text box has focus

solidcraft-cli bench-agent builds a generated ~1,300-body bicycle command by command
and times it (241 s -> 16 s here). Eval output is unchanged apart from timings.

Co-authored-by: Claude Opus 5.5 <[email protected]>

crates/doc/src/document.rs

@@ -2358,8 +2358,14 @@ impl Document {
 
     /// Evaluate an expression against the document's parameters.
     pub fn eval(&self, e: &str, kind: Kind) -> Result<f64> {
-        let (vals, _) = self.param_values();
-        Self::eval_in(&vals, e, kind)
+        // The parameters are evaluated only when the expression names one: most expressions an
+        // agent sends are plain values, and evaluating every parameter of a large design for
+        // each of them made every new feature cost more than the last.
+        let vals: std::cell::OnceCell<std::collections::BTreeMap<String, expr::Value>> = std::cell::OnceCell::new();
+        expr::eval_with(e, &|n| {
+            vals.get_or_init(|| self.param_values().0).get(n).copied().ok_or_else(|| DocError::Expr(format!("unknown parameter `{n}`")))
+        })?
+        .to_kind(kind)
     }
 
     pub fn eval_in(vals: &std::collections::BTreeMap<String, expr::Value>, e: &str, kind: Kind) -> Result<f64> {