Data should not rely on user input

Fields like `IfcExportAs`, require the user to type in the correct value, which places a huge amount of onus on the users to have a deep understanding of IFC. This add-on should provide a graphical selection method, whether its a dropdown, hierarchical tree, or autocomplete, or other, to mitigate the issue of users typing in the wrong data.

Having incorrect data typed in leads to a lot of wasted time and lots of repeated user education being required.

It can also lead to _incorrect_ data, for instance if a classification code has `[Foo]Bar:One`, and another has `[Foo]Bar:Two` in it, then two classification references will be created with the same identifier, which is incorrect.

Another example is for geolocation, where you cannot select from the EPSG registry.

No files, tests, or entry points are named. Start by locating the UI that handles fields such as IfcExportAs, classification codes, and geolocation; compare possible dropdown, hierarchical-tree, or autocomplete approaches. Done should prevent invalid or ambiguous typed values and support selecting valid classification and EPSG registry entries.

Written by the indexing model from the issue text.