core: AOT context image — init-cached recovers the bootstrap cost across processes

init in compile mode is ~2.4 s (tier loading, analyzer self-compile, macro
recompilation), paid by every process that builds a ctx from source — each
jpm-test file, embedders, workers. init-cached marshals the built ctx to a
disk image (same root-env dicts as snapshot/fork) and later processes
unmarshal it in ~5 ms, any process: nothing from the baking process is
needed at load.

The cache key fingerprints the embedded .clj stdlib (which covers jolt-core:
analyzer, IR, core tiers), the .janet seed sources next to the module, the
janet version, the init opts, and the env knobs that shape a ctx (JOLT_PATH/
MUTABLE/AOT_CORE/FEATURES) — any change rebuilds. Corrupt or non-ctx images
fall back to a rebuild (unmarshal of garbage can 'succeed' with a scalar, so
the shape is checked, not just the throw). Writes are atomic (tmp + rename)
so racing cold starts never publish a torn image. JOLT_NO_IMAGE_CACHE=1
opts out; JOLT_IMAGE_CACHE_DIR overrides the location (default TMPDIR).

Test consumers switch to init-cached (harness, suite-worker, conformance,
the behavioral unit/integration tests); tests that validate the bootstrap
itself (bootstrap-fixpoint, staged-bootstrap, aot round-trip, direct-linking)
and the deps tests (tmp-dir :paths would fragment the key) keep real init.
Full jpm test: 2:46 -> 1:58 (~29%). New ctx-image-test covers cold/warm,
cross-process load (subprocess runs defn/redef/macros/protocols/multimethods
off the baked image), per-opts keying, and corrupt-image fallback.

test/integration/self-host-test.janet

@@ -10,7 +10,7 @@
 (defn ce [ctx s] (normalize-pvecs (backend/compile-and-eval ctx (parse-string s))))
 
 (print "self-host pipeline (Clojure analyzer -> IR -> Janet)...")
-(let [ctx (init)]
+(let [ctx (init-cached)]
   # primitives + control flow
   (assert (= 3 (ce ctx "(+ 1 2)")) "+")
   (assert (= 6 (ce ctx "(* 2 3)")) "*")
@@ -58,7 +58,7 @@
 # compile path. Forms the analyzer can't handle (stateful / destructuring) fall
 # back to the interpreter, with the same observable results.
 (print "self-host via eval-toplevel routing...")
-(let [ctx (init {:compile? true})]
+(let [ctx (init-cached {:compile? true})]
   (defn ev [s] (normalize-pvecs (eval-one ctx (parse-string s))))
   (assert (= 3 (ev "(+ 1 2)")) "tl +")
   (ev "(defn sq [x] (* x x))")                 # def via self-host
@@ -70,7 +70,7 @@
   # Proof the self-hosted pipeline actually ran: only backend/ensure-analyzer
   # populates jolt.analyzer. An interpret-only ctx never loads it.
   (assert (pos? (length ((ctx-find-ns ctx "jolt.analyzer") :mappings))) "analyzer loaded under :compile?"))
-(let [ctx (init {})]
+(let [ctx (init-cached {})]
   (eval-one ctx (parse-string "(+ 1 2)"))
   (assert (zero? (length ((ctx-find-ns ctx "jolt.analyzer") :mappings))) "analyzer NOT loaded when interpreting"))
 
@@ -78,7 +78,7 @@
 # load into clojure.core at init and work the same compiled or interpreted.
 (print "clojure.core overlay (Clojure-defined core fns)...")
 (each opts [{:compile? true} {}]
-  (let [ctx (init opts)]
+  (let [ctx (init-cached opts)]
     (defn ev [s] (normalize-pvecs (eval-one ctx (parse-string s))))
     (assert (= 1 (ev "(ffirst [[1 2] [3 4]])")) "ffirst")
     (assert (= [2] (ev "(nfirst [[1 2] [3 4]])")) "nfirst")