load_checkpoint.sh v2.9:
* Added support for replaying parallel checkpoint directories.
* Changed order of operations so the replica service is started
before the (optional) replay to the offline_db.
* Fixed exit code bug where 'EXIT_CODE: 0' could be displayed
even if there were errors.
* Deprecated '-k' option to keep databases. Keeping metadata
is now the default behavior. Added '-R' option to remove metadata.
* Added a disk space safety check, projecting need based on size
of previously existing db.* files. Added a '-F' safety factor
option to control or bypass the safety.
* Fixed some errors that could be encontered in some conditions
when moving aside existing db.* and other files.
* Enhanced audit trail and generally nicer output.
* The server.locks directory is now removed.
* The rdb.lbr table and server.pid files are now handled (moved or
removed depending on -R, along db.* and other files).
* Various doc improvements.
* Added doc example, loading a parallel checkpoint on an edge.
* Added doc tip re: the common practice of creating a log file
name 'load.log' for non-interactive usage.
* Enhanced so logic to infer the type tag, which had been applied
to ServerID's specified with '-s', is now applied even if '-s' is
not used. This results in less need to specify '-l' (to skip the
license check) on server types like edges that don't need a license.
* General internal improvements, moving toward the (not yet
published) SDP Bash coding standard.