perf(lint): end File's pointer block on a slice, not a bare pointer

Round-2 code review's efficiency angle re-ran fieldalignment after
this PR's File reorder settled and found a residual, smaller finding
beyond the original 688->640 size fix: "struct with 488 pointer
bytes could be 472". RunCache *RunCache sat last among the
pointer-bearing fields; a bare pointer's entire width is its GC
pointer word, so ending the block there left no free non-pointer
tail. Reordering so linkRefs []Reference is the last pointer-bearing
field instead lets its 16-byte len/cap tail fall outside ptrdata for
free, since a slice only GC-scans its 8-byte data-pointer word.

Ref: docs/development/high-performance-go.md#struct-layout

Author: claude

internal/lint/file.go

@@ -149,15 +149,6 @@ type File struct {
 	// rather than via NewFile; LinkReferences then parses once on
 	// demand. Released once linkRefs is materialized.
 	parseCtx parser.Context
-	linkRefs []Reference
-
-	// scratch backs Memo: per-Check rule memoization. A *File is
-	// built fresh for each Check and discarded after, so values
-	// cached here never outlive a single Check — no cross-file or
-	// cross-run staleness, the same scope as the cross-file rule's
-	// per-Check cache. sync.Map keeps it safe for the concurrent
-	// readers the LSP may run against one document.
-	scratch sync.Map
 
 	// RunCache is the engine-owned read cache shared by every File
 	// processed in one engine.Run pass. Catalog and include rules
@@ -175,6 +166,23 @@ type File struct {
 	// they belong with File anyway. See plan/224.
 	RunCache *RunCache
 
+	// scratch backs Memo: per-Check rule memoization. A *File is
+	// built fresh for each Check and discarded after, so values
+	// cached here never outlive a single Check — no cross-file or
+	// cross-run staleness, the same scope as the cross-file rule's
+	// per-Check cache. sync.Map keeps it safe for the concurrent
+	// readers the LSP may run against one document.
+	scratch sync.Map
+
+	// linkRefs is declared last among the pointer-bearing fields on
+	// purpose: as a slice, only its 8-byte data-pointer word is
+	// GC-scanned when nothing after it holds a pointer, so its
+	// trailing len/cap words (16 bytes) fall outside ptrdata for
+	// free — unlike ending on a bare pointer field (e.g. RunCache),
+	// whose single word carries no such free tail. See parseCtx's
+	// comment above for what linkRefs caches.
+	linkRefs []Reference
+
 	// --- lazy-init guards -------------------------------------------
 	// Every *Done atomic.Bool / *Mu sync.Mutex pair below guards the
 	// same-named cache field above; see that field's comment for what