Better be fast for closers are closing the question.
I have been in the GIS / spatial database business for many years and have never really had my own data. Data I found useful - or work on - some of my colleagues also find interest in, in that case we use a company PostgreSQL / PostGIS, so data are seldom stored on a local machine. When I work on a heavy data analysis, I use a localhost PostgreSQL for doing the queries, since I want 100% control and the result are passed on to others and need not to be stored. In fact I never use a GIS desktop anymore for analysis and querying only for printing a map, if that is a need, a database will do all the analysis faster and better and with more reuse-ability using SQL and indexes.
A lot of file based data storage are useful also like like shape files, geopackages, spatialite, etc. If you use open source and not ready for a full blown database server I will go with SpatiaLite or a Geopackages and perhaps shape files for exchanging data with primitive users.
It is not only how your store the data, but also how you make it available to other users and your self. Your folder system for thematic data is a good abstraction. This can be made even more abstract using the QGIS QLR Browser plugin or just browse your tables in groups with styles from a QGIS project file QGZ / QGS. ArcMap does this with proprietary geodatabases also but not as fancy as QGIS. A truelig abstracted gui presents layers for you where you or others user don't need to be aware of the data source.
I seen no need of a localhost Geoserver unless you have a production one running or need some of the special Geoserver styling or are a web developer. A note is that the combination Geoserver and Open Layers produces some of the best looking rendering of geometries I have ever seen.