The Delete processing is intended to recognise when a record in the target dataset is not present in the source dataset, and take some action on the target record (usually setting the MarkAsDelete flag, as you are doing). So if, say, a discovery solution finds CIs A,B,C on day, and these are imported correctly, then the next time it only finds A,B the Delete option will set the MarkAsDelete flag in the target dataset for CI C since it seems like it no longer exists.
So if EVERY CI in your SCCM dataset exists in the target dataset, there is no Delete action required/performed. I think this is what you are seeing.
If you want to set the MarkAsDelete attribute when the SSCM data tells you you have an Inactive status, then you need to have an explicit mapping in the main data mapping section that sets MarkAsDelete to Yes only when the SCCM statsu flag says Inactive.
Hope that helps.