Regarding point 9.: I can't speak for HANA, but for Oracle it is the other way around: For best performance, do as much work as possible in (one) SQL instead breaking a query into smaller chunks in PLSQL code.
If you execute a lot of small SQLs and ...
Yes, using ROWID is not a big NO NO, as there are good reasons to use it, for example to achieve better performance avoiding an index scan in merge statements like this:
merge into xxen_report_pivot_v_recs xrpvr0 using (
select x.* from (
select
xrp...
