If you want to upload a dataset to OSM, be careful! that's called an import, and can be tricky to do. As you've noticed, it's easy to add the wrong data the wrong way. Imports are advanced topics https://wiki.openstreetmap.org/wiki/Import
OSM can store all sorts of data, so there might be a need for the "State Suburbs", esp if you want to do statistical work. But you point out that they shouldn't be used for geocoding (address lookups), so maybe those state suburbs are tagged wrong.
What areas are you looking at.
One place to look at is Mirboo. I've just picked it to avoid yielding too much PII, but it barely respects the existence of surrounding rural districts. (It might even swamp them like others I know - I'm afraid I don't have my full system here.) I know of other examples (locations to which I have an attachment), but you'll forgive me if I avoid giving too much PII on a public forum.
I'm afraid I can't give an exhaustive list because I haven't produced one. Also because I anticipate that if I produce a report it will come to nothing - I usually find that public databases are curated by people who will refuse to accept that any major corrections are necessary because they've already worked on the problem for years and therefore a major correction is taken to be an insult.
As an example, if I can only produce examples in rural areas, I fear that someone will say "we have thousands of correct addresses, but there's only twenty addresses affected by this example, and thirty by that example - it's not important enough". At that point it seems there's no point even producing a report. OSM will just fall into the category of useless databases, where, like Wikipedia, accuracy and convenience are not valued.
The OSM Wiki contains a Data Catalogue for Australia. https://wiki.openstreetmap.org/wiki/Australian_Data_Catalogu...
The issue right now is that both the PSMA data (that you want OSM to use) and the ASGS data (that you infer it is using - and in some cases that might be the source of the boundaries that were entered, but it is not officially in use by OSM) are licensed under CC-BY, which is not 100% compatible with the ODbL that OSM uses.
The necessary solution is to request an explicit waiver from the data provider, which has been requested for PSMA, but (as far as I know) never received a response.
In Australia, addresses are in the form [Street Number and Name], [Locality (i.e. suburb, town, rural district), Postcode State]. Localities are therefore very important. They need to be in any mapping database of Australia, because people will enter addresses in that form because it's what the post office uses and the local councils define.
But it looks like someone went to the Australian Bureau of Statistics and downloaded the "State Suburb" boundaries and uploaded them into Open Street Map. "State Suburbs" are not places and not used in addresses. They're statistical regions used for reporting statistics. "State suburbs" are built up of other statistical regions to approximate towns, suburbs and rural districts. They sometimes correspond well enough at a broad level to suburbs in metropolitan areas, but even in densely populated urban areas they can be wrong because what makes statistical sense isn't always the administrative border. In country areas, they take scant regard for locality boundaries. "State Suburbs" should never have been uploaded to the database.
(There's an analagous problem with Local Government Areas, which are statistical regions, and municipal districts, which are the territory of a local council. But here the distinction is going to be rarely significant because LGA boundaries are deliberately aligned to municipal district boundaries. Consequently it's more a matter of the source of the authority than the content of the data.)
You can go to data.gov.au and download locality boundaries for each state. These are what is wanted.
But I have no idea how to do this. I don't really know where to begin. I once asked someone for help about trying to find the right forum to ask for help. But it's daunting. When these online databases contain work that is wrong, it's so hard to fix them. People will be possessive of the work they've done. It's possible some of them have been corrected. Others might have been partially corrected. I just don't see how I can go through and fix them without upsetting anyone.
The alternative seems to be to just accept that there's wrong data in there because there's way too many to fix manually, even if I start now and keep going in all my spare time for the rest of my life.