Support multiple concurrent pg_basebackup backups.

With this patch, pg_basebackup doesn't write a backup_label file in the
data directory, so it doesn't interfere with a pg_start/stop_backup() based
backup anymore. backup_label is still included in the backup, but it is
injected directly into the tar stream.

Heikki Linnakangas, reviewed by Fujii Masao and Magnus Hagander.

Author: hlinnaka

doc/src/sgml/backup.sgml

@@ -814,12 +814,13 @@ SELECT pg_stop_backup();
 
    <para>
     You can also use the <xref linkend="app-pgbasebackup"> tool to take
-    the backup, instead of manually copying the files. This tool will take
-    care of the <function>pg_start_backup()</>, copy and
+    the backup, instead of manually copying the files. This tool will do
+    the equivalent of <function>pg_start_backup()</>, copy and
     <function>pg_stop_backup()</> steps automatically, and transfers the
     backup over a regular <productname>PostgreSQL</productname> connection
     using the replication protocol, instead of requiring filesystem level
-    access.
+    access. pg_basebackup does not interfere with filesystem level backups
+    taken using <function>pg_start_backup()</>/<function>pg_stop_backup()</>.
    </para>
 
    <para>

doc/src/sgml/ref/pg_basebackup.sgml

@@ -59,10 +59,9 @@ PostgreSQL documentation
   </para>
 
   <para>
-   Only one backup can be concurrently active in
-   <productname>PostgreSQL</productname>, meaning that only one instance of
-   <application>pg_basebackup</application> can run at the same time
-   against a single database cluster.
+   There can be multiple pg_basebackups running at the same time, but it is
+   better from a performance point of view to take only one backup, and copy
+   the the result.
   </para>
  </refsect1>