tree-wide: "<n>bit" → "<n>-bit"

In some places, "<n> bits" is used when more appropriate.

src/basic/user-util.h

@@ -69,13 +69,13 @@ int take_etc_passwd_lock(const char *root);
 
 /* If REMOUNT_IDMAPPING_HOST_ROOT is set for remount_idmap() we'll include a mapping here that maps the host
  * root user accessing the idmapped mount to the this user ID on the backing fs. This is the last valid UID in
- * the *signed* 32bit range. You might wonder why precisely use this specific UID for this purpose? Well, we
+ * the *signed* 32-bit range. You might wonder why precisely use this specific UID for this purpose? Well, we
  * definitely cannot use the first 0…65536 UIDs for that, since in most cases that's precisely the file range
  * we intend to map to some high UID range, and since UID mappings have to be bijective we thus cannot use
- * them at all. Furthermore the UID range beyond INT32_MAX (i.e. the range above the signed 32bit range) is
+ * them at all. Furthermore the UID range beyond INT32_MAX (i.e. the range above the signed 32-bit range) is
  * icky, since many APIs cannot use it (example: setfsuid() returns the old UID as signed integer). Following
- * our usual logic of assigning a 16bit UID range to each container, so that the upper 16bit of a 32bit UID
- * value indicate kind of a "container ID" and the lower 16bit map directly to the intended user you can read
+ * our usual logic of assigning a 16-bit UID range to each container, so that the upper 16-bit of a 32-bit UID
+ * value indicate kind of a "container ID" and the lower 16-bit map directly to the intended user you can read
  * this specific UID as the "nobody" user of the container with ID 0x7FFF, which is kinda nice. */
 #define UID_MAPPED_ROOT ((uid_t) (INT32_MAX-1))
 #define GID_MAPPED_ROOT ((gid_t) (INT32_MAX-1))