feat(ext4): enable inline_data on rootfs mkfs

Tiny files (<~160 B with default 256 B inodes) now live inside the
inode instead of getting their own 4 KiB data block. Reduces rootfs
diff size for small-file churn (pidfiles, /run state, utmp, package
metadata) and improves locality since dirty blocks cluster on
inode-table blocks rather than scattering across the data region —
fewer GCS chunks per cold resume.

mkfs-time only; applies to newly-built bases. Existing diff/discard
behaviour unchanged for files above the inline threshold.

Author: ValentaTomas

packages/orchestrator/pkg/template/build/core/filesystem/ext4.go

@@ -54,6 +54,8 @@ func Make(ctx context.Context, rootfsPath string, sizeMb int64, blockSize int64)
 			"filetype",
 			"flex_bg",
 			"huge_file",
+			// Pack file data <~160 B inside the inode to avoid a 4 KiB data block per tiny file.
+			"inline_data",
 			"large_file",
 			"sparse_super2",
 		}, ","),