installCSMDSCLI.sh env var CSM_ProductionRoot split into 2 vars

In zOS CSM DSCLI, one must run the SAMPLIB(IWNINST2) job to actually install the product via script installCSMDSCLI.sh. The script requires 3 environment variables to be set:
CSM_InstallRoot
CSM_ProductionRoot
CSM_JAVA_DIR

The CSM_ProductionRoot pathname, however, is used in two contexts:
1) As a working pathname for the script itself, in various cp, mv, rm commands;
2) As a literal text string to be sed'd into several files including the dscli executable, plus lib/CLI.CFG.

This means that the literal text used in both dscli and CLI.CFG will reflect my SMPE environment's working USS mounts and not those of the actual production home when it is ultimately deployed across my sysplexes.

Because of the nature of zOS product and/or sysres deployment, those 2 paths will likely not necessarily be the same: the file tree in which the customer executes installCSMDSCLI.sh would be different from the one in which the extracted utility is actually executed.

This requires the customer to fix the definitions in the dscli and CLI.CFG files after the installCSMDSCLI.sh script has run to reflect the actual production home of CSMDSCLI.

The CSM_ProductRoot envvar should be split into 2 variables and installCSMDSCLI.sh updated to reference the correct one based on the context of use: either the working install directory, or the literal pathnames which get inserted into dscli and CLI.CFG.

For instance, I install CSMDSCLI in a separate directory tree in SMPE as:
/Service/zos/opt/IBM/CSMDSCLI
But I would run it out of here on my live sysres:
/opt/IBM/CSMDSCLI

However, installCSMDSCLI.sh uses the same CSM_ProductionRoot variable to populate those path names.

The current installCSMDSCLI.sh does not differentiate between those 2 contexts and now I must correct them manually on my end whenever I apply maintenance and rerun installCSMDSCLI.sh.