This PR adds the filter option for tags and attic data to the overpass request as well as enabling custom overpass instances. In addition the CLI is refactored to make way for a python API (function parameters) and also make use of argparse. Splitting the parser creation and the argument augmentation into different functions will enable users to add these arguments to their own parser. The PR otherwise aims to leave the original code as untouched as possible.

closes #1 and closes #2

To make the tool more easily usable in another python module/package the CLI could be separated from the rest of the code by creating a python-usable function that takes bbox, delete?::boolean and authentication as input.

Sometimes only a subset of the large OSM Data is relevant for a tester inside the sandbox. It would be great if one could specify a filter (of tags) that should be transferred to the sandbox. Having other attributes (like e.g. the https://docs.ohsome.org/ohsome-api/v1/filter.html would be great, but I don't think its necessary here).

This PR

• makes the recursive overpass query more explicit
• makes the sandbox an exact copy of the current OSM data by
  • always cleaning the sandbox,
  • allowing empty uploads (meaning that the sandbox stays empty just as the OSM database is at that location),
  • skipping any data filtering.

Especially the last point may be a question of taste, so feel free to comment or close the PR, I just wanted to share the edits I make in my local version.filter_by_bbox may be better done in overpass by including line 239/240 (as it was before). Yet this will destroy larger objects. delete_missing' may prevent upload errors but should not occur for the recursive query. So this way users will be informed about broken data instead of having it silently removed.delete_unreferenced_nodes` is a cleaning step that is not in line of making the sandbox a 1:1 copy.