p2: digest with in-sift FIFO, SQLite caches, Honcho context, recover

Dixie is capped (max_inflight 2, queue_depth 8, queue_wait 20s) and
never queued on llama-server. Streaming chat, thinking off, 45s
fallback to clean. Digest and chunk-embedding caches are
content-addressed SQLite (Honcho context is not cached). Recover
uses the guarded Fetcher. guardian-star is refused in code.

vendor/modernc.org/libc/mem_expvar.go

@@ -0,0 +1,12 @@
+//go:build libc.memexpvar
+
+package libc
+
+import "expvar"
+
+func init() {
+	// make sure to build with -tags=memory.counters to have the actual data accumulated in memory allocator
+	expvar.Publish("memory.allocator", expvar.Func(func() interface{} {
+		return MemStat()
+	}))
+}