perf: build queries up to 2.7x faster by reusing ones already built

docs/compat.md

@@ -70,7 +70,7 @@ These are the cases where the proof holds Python to something narrower than "ide
 - **Interned queries.** Upstream's builder hands back the query it built before when a step repeats one,
   keyed so that some values JavaScript's types rule out share a key: a `flip` of `'yes'` with `true`, any
   `orderBy` direction but `'asc'` with `'desc'`. In one process, such a value can come back as whichever
-  was built first. Python builds what it is given. The harness gives each program a fresh schema, which
+  was built first. Python interns as well, but by exact keys, so it builds what it is given. The harness gives each program a fresh schema, which
   upstream interns by, so that no comparison depends on the programs before it.
 - **A mutator returning `null`.** Upstream records a mutator's return value as result data unless it is
   `undefined`, so a TypeScript mutator can return `null` as data. Python has only `None`, which the port

src/zero_server/protocol/ast.py

@@ -11,7 +11,7 @@
 
 import functools
 from collections.abc import Callable, Sequence
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from typing import TYPE_CHECKING, Any, ClassVar, Literal, Protocol
 
 from zero_server.protocol.wire import dump_json, json_model, parse_wire
@@ -192,6 +192,9 @@ class CorrelatedSubquery:
     hidden: bool | None = None
     subquery: AST
     system: System | None = None
+    # The builder query `subquery` is the AST of, kept alive with it: the builder interns a query by
+    # the identity of its subqueries' ASTs, so they must outlive the query keyed by them.
+    source: Any = field(default=None, repr=False, compare=False)
 
     def to_json(self) -> dict[str, Any]:
         out: dict[str, Any] = {"correlation": self.correlation.to_json()}