test(fetchguard): widen streaming-cap ceiling to stop flake under parallel runs

TestFetchTooLargeIsStreaming asserted the server wrote at most 64 KiB, but
the handler keeps flushing 64-byte chunks until the client's close propagates,
which under a loaded full-suite run exceeds that (observed ~80 KiB). The
assertion guards against unbounded buffering toward 8 MiB, so 1 MiB keeps
the intent and removes the flake.

fetchguard/fetch_test.go

@@ -183,9 +183,12 @@ func TestFetchTooLargeIsStreaming(t *testing.T) {
 	if reasonFrom(t, err) != ReasonTooLarge {
 		t.Fatalf("reason = %q, want %s", reasonFrom(t, err), ReasonTooLarge)
 	}
-	// TCP/HTTP buffering can write a little past maxBytes+1; the client must
-	// not have pulled an unbounded body first (the handler would hit 8MiB).
-	if got := written.Load(); got > 64<<10 {
+	// TCP/HTTP buffering can write past maxBytes+1, and under a loaded parallel
+	// test run the handler keeps flushing 64-byte chunks until the client's
+	// close propagates (observed ~80 KiB). The client must not have pulled an
+	// unbounded body first (the handler would hit 8 MiB), so 1 MiB is the
+	// meaningful ceiling.
+	if got := written.Load(); got > 1<<20 {
 		t.Fatalf("server wrote %d bytes, client appears to have buffered unbounded body", got)
 	}
 	if written.Load() < maxBytes+1 {